1. Introduction

As a non-invasive imaging technology with rapid feedback, high resolution, and no ionizing radiation, luminescent bioimaging is widely used in cancer detection, vascular visualization, and intraoperative guided surgery [1–3]. Luminescent bioprobes, like organic dyes, quantum dots (QDs), and single-walled carbon nanotubes (SWCNTs), are competitive candidates for bioimaging to reveal the physiological and pathological processes of organisms. Among them, organic dyes have been widely studied due to their easy design of optical properties and good biocompatibility. However, the easy photobleaching, poor stability, and inability to long-term imaging of organic dyes limit their further clinical applications [4]. For QDs, they have high stability and luminescent quantum yield. Nevertheless, heavy metal compositions (e.g. Hg, Cd, and Pb) in most QDs make their safety of biological applications controversial [5]. Although SWCNTs show wide absorption bands (UV, visible, and NIR regions), the luminescent quantum yield remains quite low (usually less than 3%), and the needle-like structure with typical size of several hundred nm may cause tissue damage [6,7].

Alternatively, luminescent bioprobes using rare earth doped nanoparticles (RENPs) have attracted increasing attention due to their advantages of ultrahigh photostability and chemical stability [1,8], long luminescence lifetime (up to the order of ms) [9], high biocompatibility [10,11], and large Stokes and anti-Stokes shift [8]. More importantly, unlike the commonly used luminescent bioprobes being excited by UV or visible light and emitting visible light with longer wavelength [12], RENPs-based luminescent bioprobes exhibit superior penetration ability in biological tissues owing to their flexible excitation and emission wavelengths, originating from the abundant ladder-like energy levels of RE ions [13]. The flexible excitation and emission wavelengths of RENPs enable people to utilize the low-loss excitation and/or emission bands falling in the so-called biological windows (BWs), and to achieve deeper penetration of the luminescent probes [14].

The BWs are located in the NIR wavelength region, comprising the NIR-I BW within 0.7-0.95 µm, the NIR-II BW ranging from 1 to 1.35 µm, and the NIR-III BW of 1.55-1.87 µm [15]. Clearly, manipulation of the excitation and emission light of RENPs probes both falling in the BWs not only avoids the autoluminescence caused by UV or visible light excitation, but also reduces the light absorption and scattering of biological tissues. As a result, these NIR-NIR probes can achieve high imaging signal-to-noise ratio (SNR) and improved penetration depth in tissues.

The early-stage NIR-NIR RENPs probes focused mainly on the NIR-I and NIR-II BWs. In 2008, Nyk et al. reported Tm³⁺ based NIR-I bioprobes (excitation at 0.98 µm and emission at 0.8 µm), showing deep-tissue imaging with high SNR [16]. Later on, they further developed alpha-phase Tm³⁺ based core-shell (C-S) bioprobes operated in the NIR-I BW, capable of whole-body imaging of mouse with a SNR of 310 and ex vivo bioimaging with penetration depth of cm level [17]. With the even better penetration ability in biological tissues, NIR-II bioprobes are attracting increasing attention. In 2016, Jiang et al. designed LiYF₄:Nd³⁺ bioprobe with bright luminescence at 1.05 µm under 0.8 µm excitation and utilized them for in vivo vascular NIR-II imaging with high SNR [18]. Also using Nd³⁺ activator, Villa et al. constructed SrF₂:Nd³⁺ bioprobe to realize intense 1.34 µm luminescence, which can avoid the autoluminescence from the diet of specimen and achieve high contrast in vivo bioimaging [19].

Compared with the bioprobes operated in NIR-I and NIR-II BWs, researchers recently revealed that the NIR-III bioprobes may achieve even higher SNR and spatial resolution in deep-tissue imaging due to the further reduced scattering loss and the suppressed tissue autoluminescence [20]. In addition, NIR-III bioprobes based on the down-conversion (DC) RENPs have relatively high quantum yields, enabling the production of strong enough luminescence for biological imaging upon low excitation light intensity [21]. In 2013, Naczynski et al. reported the NIR-III bioprobe based on NaYF₄:Yb³⁺,Er³⁺@NaYF₄ C-S NPs, and firstly revealed the improved penetration depth and resolution [22]; In 2017, Zhong et al. proposed a NIR-III bioprobe based on NaYbF₄:Er³⁺,Ce@NaYF₄ C-S NPs, and further demonstrated the fast imaging of mouse cerebral vasculatures using the improved DC intensity of Er³⁺ 1.5 µm emission [23]. In 2018, Wang et al. used Er³⁺ heavily doped C-S NPs as bioprobe to achieve ex vivo HeLa cell imaging and mouse brain imaging under 0.8 µm excitation [24]. To date, the NIR-III bioprobes based on RENPs utilize mainly the 1.5 µm emission of Er³⁺, and typically require 0.8 or 0.98 µm excitation. However, 0.98 µm excitation leads to the concerns of overheating of biological tissues and thus the poor penetration depth [25], while the relatively weak absorption of Er³⁺ at 0.8 µm is not conducive to generate high brightness luminescence [26].

In addition, although larger than that of the traditional luminescent bioprobes, the Stokes shift of current RENPs based bioprobes is still limited, causing the spectra overlap and thus would restrain the multiplexed bioimaging and labeling applications. Very recently, Wang et al. constructed a NIR-III bioprobe based on molecular Er³⁺-bacteriochlorin complexes with a large Stokes shift of ∼0.76 µm, which renders a powerful tool for visualizing of dynamic circulatory, metabolic processes, and tracking of cancer cell metastases in mouse brain [27].

Herein, we develop a novel 0.8 µm laser excited NIR-III bioprobe with efficient 1.8 µm emission, based on Tm³⁺ heavily doped NaYF₄ NPs. Combining the stronger 0.8 µm absorption of Tm³⁺ as compared to that of Er³⁺ [26], the efficient cross relaxation (CR) process, and the effective inert shell passivation, the 1.8 µm emission intensity of Tm³⁺ is greatly enhanced. Compared with the conventional NIR-III bioprobes using Er³⁺ 1.5 µm emission, Tm³⁺ 1.8 µm emission shows weaker light scattering and autoluminescence in tissues [28] and a large Stoke shift of ∼1 µm, which make the as-prepared Tm³⁺ based NIR-III bioprobe a competitive candidate for deep-tissue applications. After investigation the luminescence mechanisms, we further assess the penetration ability of the newly developed bioprobe, by ex vivo comparison experiments. It is found that the as-prepared NIR-III bioprobe can achieve deep-tissue imaging while avoiding the problem of laser overheating effect, which can find wide applications including disease diagnosis, high-resolution bioimaging, in vivo temperature feedback, etc.

2. Experimental

The hexagonal NaYF₄ core-only NPs were prepared via a slightly modified literature procedure [29]. 1 M YCl₃, 0.5 M NaOH and NH₄F methanol solutions were prepared firstly. Then, 1 mL of YCl₃ methanol solution, 10 mL of oleic acid (OA), and 15 mL of 1-octadecene (ODE) were added to a 50 mL three-necked flask. The mixture was heated at 180 °C with stir for 45 min before cooling down to 50 °C. Subsequently, 5 mL of NaOH and 8 mL of NH₄F methanol solution were added into the mixture and stirred for another 45 min. After the reaction mixture was heated at 110 °C for 15 min to remove the residual methanol and water, the solution was quickly heated to 300 °C and kept for 1.5 h before cooling down to room temperature. The as-prepared NPs were precipitated by addition of ethanol, collected by centrifugation at 8000 rpm for 5 min, and washed with ethanol and methanol for several times. The final product was dispersed in 10 mL cyclohexane. Argon gas was adopted throughout the entire experiment to protect the reaction.

The as-prepared NaYF₄ core-only NPs were used as seeds for shell modification. In a typical experiment, the shell solution was first prepared by mixing x/100 mL (x = 0.5, 25, and 100) of TmCl₃ methanol solution (1 M), OA (5 mL), and ODE (7.5 mL) in a 50 mL three-necked flask. The resulting mixture was heated at 180 °C for 45 min before cooling down to 50 °C. 2.5 mL of NaOH and 4 mL of NH₄F methanol solutions and 5 mL of NaYF₄ seed solution were then added and stirred for 45 min. After heating at 110 °C for 15 min to remove the residual methanol and water, the reaction mixture was quickly heated and kept at 290 °C for 1.5 h. The resulting C-S NPs were washed and dispersed in cyclohexane following the above mentioned route. Argon gas was used throughout the entire experiment to protect the reaction. The as-prepared C-S NPs were denoted as Y@xTm NPs (x = 0.5, 25, and 100 mol.%) thereafter.

Finally, the Y@xTm NPs were again used as seeds for another shell modification, following a similar route. The two differences are 0.5 mL LuCl₃ methanol solution (1 M) was used in the shell solution, and 4 mL of Y@xTm cyclohexane dispersion was added as the seeds. The resulting the C-S-S NPs were denoted as Y@xTm@Lu NPs thereafter.

The morphologies of the samples were recorded via a transmission electron microscopy (TEM, Tecnai G2, FEI). Room temperature luminescence measurements were performed by irradiating the samples via a variable-power NIR diode laser (LWIRL800-5W, Laserwave), and the spectra were measured by a customized ultraviolet to mid-infrared spectrometer (QM8000, Horiba). The InGaAs detector was equipped with a lock-in amplifier. The time-resolved photoluminescence curves were recorded by the same spectrometer, using a signal generator (DG1032Z, Rigol) to modulate the laser output from continuous-wave (CW) into square-wave.

3. Results and discussion

3.1 Design of the Tm³⁺ heavily doped bioprobe

Besides the commonly used Er³⁺ in the RENPs bioprobes, Tm³⁺ also shows considerable potentials as bioprobes owing to its unique electron configuration. The conventional Tm³⁺ activated bioprobes usually emit NIR-I photons at 0.8 µm (³H₄ → ³H₆) under 0.98 µm excitation, with the sensitization of Yb³⁺ [30–32]. As for the NIR-III bioprobes, Tm³⁺ can be excited at 1.2 µm, and then enables the DC luminescence of Er³⁺ (1.5 µm) through the energy transfer (ET) from Tm³⁺ to Er³⁺ [33]. Another case of NIR-III bioprobe containing Tm³⁺ was reported in Er³⁺ heavily doped NPs, where Tm³⁺ was used as the energy trapping centers to suppress the inner quenching, which is induced by the deleterious CR, the phonon-assisted ET, and the energy trapping by the inner lattice defects [12], and therefore can help the generation of strong Er³⁺ 1.5 µm emission [34]. Clearly, compared with the important roles of Er³⁺ played in the NIR-III bioprobes, Tm³⁺ is rarely used as the NIR-III activator, although it can generate luminescence band at 1.8 µm.

As depicted in Fig. 1(a), we design a Tm³⁺ activated NIR-III bioprobe, using Tm³⁺ heavily doped NaYF₄ as the luminescent layer and further coated with a NaLuF₄ inert shell. In this design, Tm³⁺ not only acts as sensitizer to harvest the incident light energy, but also serves as activator to emit the NIR luminescence at 1.47 and 1.8 µm under 0.8 µm excitation. Specifically, electrons in the ground state ³H₆ are firstly pumped to the high-lying ³H₄ state after absorbing an incident photon. Then, these excited electrons can feed the lower-lying states radiatively and nonradiatively via the following processes: Nonradiative decay of ³H₄ → ³H₅ + phonon (∼4300 cm⁻¹), two radiative transitions of ³H₄ → ³F₄ + photon (1.47 µm) and ³H₄ → ³H₆ + photon (0.8 µm), and CR of ³H₄ + ³H₆ → ³F₄ + ³F₄ between neighboring Tm³⁺. After ³F₄ state being populated by aforementioned CR, the 1.8 µm emission band is activated, correspond to the ³F₄ → ³H₆ transition (Fig. 1(b)).

 

Fig. 1. (a) Schematic of the design NIR-III bioprobe. (b) The simplified energy level diagram showing NIR-III luminescence pathways of Tm³⁺ heavily doped NPs under 0.8 µm excitation.

Download Full Size | PDF

The 1.47 µm emission band of Tm³⁺ overlaps with the absorption peak of water molecules, making this emission not conducive to biological applications [35]. But fortunately, the remaining 1.8 µm emission band arouses weak absorption and low scattering in biological tissues [28], validating the deep penetration in vivo. Therefore, this work aims mainly at building Tm³⁺ based NIR-III bioprobe that is capable of emitting highly efficient 1.8 µm luminescence upon 0.8 µm excitation. Herein, besides the conventional low doping concentration of 0.5 mol.% Tm³⁺ in RENPs, we choose much higher doping levels (25 and 100 mol.%) to trigger the intense Tm³⁺ 1.8 µm emission. On the one hand, high doping can increase the absorption of excitation light energy, and on the other hand, it can strengthen the CR between Tm³⁺, which facilitates the population of 1.8 µm energy level. However, the increase of doping concentration also results in strong surface quenching, the absorbed energy trapped by the surface quenchers (typically the surface defects, ligands, and solvent molecules), which often happens in high doping RENPs as high doping exacerbates these surface nonradiative deactivations [11,36]. Thanks to the recently developed inert passivation technique, which can effectively inhibit the surface quenching, and thus the RENPs with an inert shell can generally produce brighter luminescence. Therefore, a NaLuF₄ inert shell is required outside the Tm³⁺ heavily doped layer.

3.2 Structural characterizations of the as-prepared bioprobes

To fairly compare the luminescence signals between bioprobes with different Tm³⁺ doping concentrations, pure NaYF₄ core-only NPs were pre-synthesized as a template by a modified thermal decomposition method [29]. NaYF₄ template NPs were used as seeds for further coating shells with different Tm³⁺ doping levels, forming the Y@xTm C-S NPs. Finally, a NaLuF₄ inert shell was coated as the outermost layer, forming the Y@xTm@Lu C-S-S NPs (Fig. 2(a)). As shown in Fig. 2(b), the NaYF₄ template NPs are well mono-dispersed nanorods, with two characteristic average sizes of 15 and 18 nm. After the epitaxial growth, the resulting C-S and C-S-S NPs all remain well mono-dispersed and show uniform nanorods morphology. With the increase of Tm³⁺ doping level from 0.5 to 25 and to 100 mol.%, the mean diameters of Y@xTm NPs are 20, 22, and 22 nm, and the mean lengths are 31, 38, and 31 nm, respectively (Fig. 2(c)–2(e)). As for the Y@xTm@Lu NPs, the mean diameters are 29, 32, and 32 nm, and the mean lengths are 35, 40, and 34 nm, with the increase of Tm³⁺ doping level (Fig. 2 f–2 h). These enlarged particle sizes manifest the successful coating of different shells. The similar sizes of the resulting NPs with different Tm³⁺ doping levels enable the convincing comparisons of their luminescence properties, as the luminescence properties of RENPs are highly sensitive to the particle morphology. In addition, high-resolution TEM image of Y@100Tm@Lu NPs (inset of Fig. 2 h) displays clear lattice fringes with an observed d spacing of 0.29 nm, in good agreement with the lattice spacing of the (101) plane of hexagonal-phased NaYF₄ (JCPDS 16-0334), thus verifying the high crystallinity of the resulting RENPs.

 

Fig. 2. (a) Schematic diagrams showing the structural evolution of the core template, C-S, and C-S-S NPs. TEM images of the (b) NaYF₄ core-only template NPs, the (c) Y@0.5Tm, (d) Y@25Tm, and (e) Y@100Tm C-S NPs, and the (f) Y@0.5Tm@Lu, (g)Y@25Tm@Lu, and (h) Y@100Tm@Lu C-S-S NPs. Scale bars are all 100 nm in the TEM images. Inset in (h) shows the high-resolution TEM image of the Y@100Tm@Lu NPs.

Download Full Size | PDF

3.3 Luminescent properties of the as-prepared bioprobes

Figure 3(a)–3(c) show the DC luminescence of C-S and C-S-S NPs with different Tm³⁺ concentrations excited by 0.8 µm laser, using an excitation power density of ∼7 W/cm². As expected, two emission bands appear at 1.47 and 1.8 µm, corresponding to the radiative transitions of ³H₄ → ³F₄ and ³F₄ → ³H₆, respectively. The full width at half maximum of 1.47 µm emission band is ∼50 nm, while it increases to ∼150 nm for 1.8 µm emission band. With the increase of the concentration of Tm³⁺, the luminescence intensities of Y@xTm C-S NPs at 1.8 µm and 1.47 µm both decrease gradually (Fig. 3(d)), leading to the extremely weak luminescence intensity of Y@100Tm NPs. This can be mainly attributed to the surface quenching effect. The surface quenching effect can be alleviated by the inert shell passivation (Fig. 3(a)–3(c)), owing to the suppression of energy trapping by the surface quenchers [36]. As summarized in Fig. 3(e), after inert shell passivation, low doping C-S-S NPs (Y@0.5Tm@Lu) show slight increase in luminescence intensity as compared to their C-S precursors (Y@0.5Tm), which can stem from the weak surface quenching aroused by the diluted Tm³⁺ in the host lattice. In stark contrast, coating the inert layer significantly enhances the luminescence intensity of the heavily doped NPs (25 and 100 mol.%), especially for the 1.8 µm emission band. As shown in Fig. 3(f), the enhanced factor aroused by inert passivation for 1.47 and 1.8 µm emissions increases gradually as 2.7 → 2.1 → 2.7 and 2.0 → 41.7 → 240.4, respectively, with the increase of Tm³⁺ doping concentration. These remarkable enhancements obtained in the heavily doping C-S-S NPs as compared to their C-S counterparts can be ascribed to the associated effects of the strong energy harvest and the suppression of surface quenching, which is consistent with the previous results [37].

 

Fig. 3. NIR luminescence spectra of the C-S and C-S-S NPs doped with (a) 0.5, (b) 25, and (c) 100 mol.% Tm³⁺ under 0.8 µm excitation. The integral intensities of 1.47 and 1.8 µm emission bands of (d) C-S NPs and (e) C-S-S NPs. (f) The enhanced factors of two NIR emission bands aroused by inert shell passivation.

Download Full Size | PDF

The evidently improved 1.8 µm emissions, falling in the NIR-III BW, of Tm³⁺ heavily doped C-S-S NPs validate our previous design. In addition, a very large Stokes shift of around 1 µm for the proposed bioprobe, upon excitation at 0.8 µm and emission at 1.8 µm, is even larger than the recently reported large shift of 0.76 µm [27]. This large shift can facilitate the multiplexed bioimaging and labeling for applications of luminescence-guided surgery and diagnostics.

To validate the proposed mechanisms of Tm³⁺ NIR emission in more detail, the variation of the luminescence intensity ratio (LIR) of ³F₄ emission (1.8 µm) to ³H₄ (1.47 µm) emission with Tm³⁺ doping concentration is shown in Fig. 4(a). The LIR of C-S NPs increases first and then decreases (1.8 → 2.6 → 1.7) as Tm³⁺ concentration increases. The first increase can mainly stem from the CR process, as labeled in Fig. 1(b), which populates the 1.8 µm level and meanwhile depopulates the 1.47 µm level. With the increase of doping concentration to 100 mol.%, the surface quenching becomes dominant. It seems that the 1.8 µm emission is more sensitive to the surface quenchers as compared to 1.47 µm emission, thus 1.8 µm emission is quenched more evidently and results in a decreased LIR. The LIRs of Y@0.5Tm and Y@0.5Tm@Lu NPs showed little difference (1.8 and 1.3) due to the weak surface quenching effect. In stark contrast, the LIR of C-S-S NPs increases gradually and evidently (1.3 → 51.1 → 153.2) as Tm³⁺ concentration increases from 0.5 to 100 mol.%. This indicates that the aforementioned CR process dominates the population mechanisms in heavily doped RENPs with an inert shell passivation, which verifies experimentally our previous design.

 

Fig. 4. Evidences of the proposed mechanisms of the Tm³⁺ NIR emissions. (a) Luminescence intensity ratios of ³F₄ emission to ³H₄ emission of core-only and C-S NPs with Tm³⁺ doping concentration. (b) Luminescence decay profiles of the ³H₄ and ³F₄ emissions of different NPs.

Download Full Size | PDF

Besides, we measure the luminescence decay profiles of the Tm-based NPs at 1.47 and 1.8 µm, under 0.8 µm square-wave excitation (Fig. 4(b)). First, the prolonged decay processes for both 1.47 and 1.8 µm emissions of Y@0.5Tm@Lu NPs compared with Y@0.5Tm NPs verify that the surface quenching is reduced by the inert shell. Second, compared with Y@0.5Tm@Lu NPs, the decay processes of ³H₄ and ³F₄ levels in heavily doped (25 and 100 mol.%) NPs are much faster, indicating the existence of strong inner quenching. Third, the 1.8 µm lifetimes of different NPs are generally larger than that of 1.47 µm emission, which can be mainly attributed to the stronger multi-phonon decay of ³H₄ level as compared to that of the ³F₄ level, as ³H₄ has a smaller energy gap to its lower level.

Figure 5(a) depicts the experimental setup showing the good penetration ability of the proposed NIR-III NPs. 0.8 µm laser with an output power density of 1.0 W/cm² is focused by a lens to illuminate the bioprobes behind the chicken breast tissue. The resulting NIR-III luminescence signal passing through the chicken breast is focused and then collected by an InGaAs detector equipped on a spectrometer. Herein, besides the as-prepared Y@100Tm@Lu NPs, 80Er20Yb@Lu NPs with a reported high quantum yield (QY) of 11% representing the state-of-the-art deep penetration NIR-III bioprobe is chosen as a Ref. [24].

 

Fig. 5. Comparison experiments using the as-prepared Y@100Tm@Lu bioprobe and the state-of-the-art NIR-III bioprobe. (a) The schematic diagram of the experimental setup demonstrating the strong penetration ability of Y@100Tm@Lu NPs as bioimaging probe. Insets are the images of the chicken breast. (b) The NIR-III luminescence spectra of Y@100Tm@Lu (1.8 µm emission band) and 80Er20Yb@Lu (1.55 µm emission band) NPs passing through chicken breast tissues with thicknesses of 0, 2.1, 3.5, and 5.6 mm. Inset is the TEM image of the as-prepared 80Er20Yb@Lu NPs. (c) Integral emission intensities of two NPs passing through different chicken breast tissues.

Download Full Size | PDF

Penetration depth is critical for luminescence bioprobes, which is mainly affected by the attenuation of biological tissue, including absorption and scattering. The initial integral luminescence intensity of Tm-based bioprobe is 110% of that of Er-based one, under the identical excitation conditions, and both show very high SNR (Fig. 5(b)). The 1.8 µm emission QY of Y@100Tm@Lu bioprobe is estimated to be ∼5.6% (800 nm excitation, 7 W/cm²). Although this QY value is lower than that of the 80Er20Yb@Lu bioprobe (11%), the 100 mol.% doping level of Tm³⁺ enables stronger energy harvest, thus resulting in a brighter luminescence. As shown in Fig. 5(b)–5(c), after passing through chicken breast with thickness of only 2.1 mm, the NIR-III luminescence signals from two NPs were seriously attenuated, but still can be clearly distinguished from the background. With the thickness increasing from 2.1 to 3.5 and to 5.6 mm, the signal further gradually decreases. The changes of spectral shapes after passing through the tissue can be attributed mainly to the wavelength-dependent absorption [27]. Importantly, the luminescence signal of Tm-based bioprobe showed less energy loss, it almost doubles the signal intensity of that of the Er-based one (the ratio of I_Tm/I_Er reached ∼180%) after attenuated by a 5.6 mm chicken tissue, which is roughly the limit penetration depth of the Tm-based bioprobe under the current experimental conditions. This comparison verifies the deep-tissue penetration ability of the proposed Y@100Tm@Lu NPs.

4. Conclusion

In conclusion, NaYF₄, NaYF₄@NaYF₄:Tm³⁺ and NaYF₄@NaYF₄:Tm³⁺@NaLuF₄ NPs were prepared by a thermal decomposition method. After inert shell passivation, the NIR-III luminescence intensity of Tm³⁺ at 1.8 µm increases significantly in heavily doped NPs as compared to that in low doping NPs, which was mainly attributed to the efficient energy harvest combined with strong CR process. However, the inner quenching effect was still considerable, leading to the shorter luminescence lifetimes of heavily doped NPs. The proposed luminescence mechanisms were verified based on the variation of 1.8 to 1.47 µm LIR with the doping concentration, as well as the decay profiles of the luminescence. In addition, ex vivo experiment revealed that the penetration ability in biological tissue of the proposed Tm³⁺-based bioprobe was better as compared to that of the state-of-the-art Er³⁺-based bioprobe. The Tm³⁺ heavily doped NPs with intense 1.8 µm luminescence, large Stoke shift up to 1 µm, and deep-tissue penetration, are promising bioprobe for in vivo multiplexed imaging, luminescence-guided diagnostics and surgery.