CHAPTER 1: Lanthanide-Based Near Infrared Nanomaterials for Bioimaging
Published:05 Aug 2016
Lanthanide-based near infrared (NIR) nanomaterials, when excited by continuous-wave NIR light, exhibit a unique photoluminescence with higher or lower energy, corresponding respectively to an upconversion (UC) or downconversion (DC) process. This special luminescence property makes them promising as bioimaging probes with attractive features, such as no autofluorescence from biological samples and a large penetration depth. As a result, lanthanide-based NIR nanomaterials have emerged as novel agents for imaging of small animals. In this chapter, recent reports regarding the biomedical imaging investigation using these novel nanoprobes are summarized. Finally, we discuss the challenges and opportunities in the development of lanthanide-based NIR nanomaterials for biomedical imaging.
Lanthanide elements are spectroscopically rich species, a property that facilitates their use as optical codes in a spectral window distinct from fluorescent dyes used for labeling biological samples. The lanthanide 4f orbitals are buried beneath the 6s, 5p, and 5d orbitals; hence, spectra arising from f–f transitions are narrow and insensitive to their environment, unlike transition metal (3d) spectra.1 Most importantly, this gives rise to a rich energy-level structure in the near infrared (NIR), visible (VIS), and ultraviolet (UV) spectral range (Figure 1.1). Triply ionized lanthanide ions in solid hosts typically have emission line widths of ∼10–20 nm (FWHM, full width at half maximum), which is about half that observed for quantum dots (QDs, ∼25–40 nm) and much narrower than that observed for organic dyes (∼30–50 nm) or transition metal ions (∼100 nm).2,3 This feature allows more resolvable bands to be packed into the same spectral bandwidth, which enables a larger number of distinct combinations. Because lanthanide emissions involve only atomic transitions, they are extremely resistant to photobleaching. The energy-level structure in lanthanide ions also creates the possibility for large shifts between the excitation and emission bands. This shift can be several hundred nanometers, containing discrete gaps with zero absorption. By comparison, the HOMO–LUMO (highest occupied molecular orbital–lowest unoccupied molecular orbital) transition in organic dyes typically results in overlapping excitation and emission bands and a Stokes shift of only 10–30 nm between the absorption and emission maxima. The large variety of absorption and emission wavelengths, the independence on host materials, and low vibration energy losses make lanthanides ions be ideal for spectral conversion. Lanthanide ions can be doped in a variety of solids such as crystals, fibers, or glass ceramics to give them the desired downconversion and upconversion optical properties. Thus, their remarkable luminescence properties have been widely applied in lasers, solar cells, analytical sensors, photodynamic therapy, and optical imaging.4–6 In this chapter, we focus on some distinct characteristics of lanthanide-based NIR nanomaterials that are closely related to their bioimaging applications, such as color tunability, energy transfer principles, and some strategies for enhancing luminescent efficiency. In addition, we systematically introduce the most recent bioimaging work based on lanthanide NIR nanomaterials.
1.2 Upconversion Nanoparticles (UCNPs)
Upconversion materials, which emit high-energy photons under excitation by the NIR light (anti-Stokes shift) were first discovered in the 1960s,7 but have primarily been exploited for the development of some remarkably effective optical devices such as infrared quantum counter detectors,8,9 temperature sensors,10,11 and compact solid state lasers.12–15 Thus, for more than 30 years the use of the upconversion effect has been limited to bulk glass or crystalline materials.16–19 Because of their suitable size (small enough to go in and out through many biological host materials, such as the cytoplasm or nucleus of a cell) and their unique properties, such as high chemical stability, low cytotoxicity, and high signal-to-noise ratio, the biological applications of UCNPs for analytical assays and bioimaging were readily recognized.20–23 The upconversion process proceeds by different mechanisms: excited state absorption (ESA), energy transfer upconversion (ETU), and photon avalanche (PA).24 ESA and ETU are based on the sequential absorption of two or more photons by metastable, long-lived energy states. In the case of ESA, the ground state activator absorbs at least two photons of suitable energy sequentially. In ETU, activator and neighboring sensitizer both absorb one photon at first, then energy is transferred between sensitizer and activator, resulting in a population of emitting ions in a highly excited state.25 PA was first reported by Chivian et al. in 1979.26 They found that when Pr3+-doped LaCl3 or LaBr3 crystal is exposed to laser-pump radiation slightly in excess of a certain critical intensity, the fluorescence of Pr3+ increases by orders of magnitude.13,26 PA-induced UC features an unusual pump mechanism that requires a pump intensity above a certain threshold value and always responds slowly to excitation (up to several seconds). The quantum yield (QY) is differs considerably among these three mechanisms: in theory, ESA < ETU < PA, but PA always needs a rather high excitation energy and suffers from slow response to excitation. The QY of ETU is two orders of magnitude higher than that of ESA, making it well understood and widely applied in many fields.7
Efficient UCNPs are composed of three components: a host matrix, a sensitizer, and an activator. An ideal host matrix needs to be optically transparent and have low lattice phonon energy, in order to minimize non-radiative losses and maximize radiative emission. NaYF4,27–29 NaGdF4,30,31 NaLuF4,32 LaF3,33 and CaF2,34 among others, have proven to be ideal candidates. With the development of nanotechnology, several methods have been used to fabricate uniform monodispersed UCNPs with controlled crystalline phases and sizes, including coprecipitation,35–42 thermal decomposition,43–47 hydro(solvo)-thermal synthesis,48–62 sol–gel process,63–66 and combustion synthesis,67 which have been reviewed in many papers.68–72 A proper choice of synthesis method enables the development of UCNPs whose properties match the need for the applications envisioned. On the other hand, the rational choice of different lanthanide ions as sensitizer and activator is also very important. Typically, Yb3+ is always chosen as sensitizer because of its large absorption cross-section of around 980 nm. Er3+, Tm3+, and Ho3+ feature ladder-like arranged energy levels which favor multiphoton process, and thus are frequently used as activators.25 Besides these three components, rational design of core–shell structure to minimize surface quenching effects and improve the luminescence efficiency of UCNPs is also very crucial.25,70–72
1.2.1 UCNPs Excited at 980 nm
The upconversion materials were first used in tissue imaging in 1999, when Zijlmans and coworkers reported the first upconversion bioimaging based on submicron-sized Y2O2S:Yb3+/Tm3+ particles (0.2–0.4 µm).73 Upon excitation at 980 nm, they observed a low autofluorescence signal and no bleaching even after continuous exposure to high excitation energy levels. After that, the idea of upconversion bioimaging was realized by employing other oxysulfide or oxide nanomaterials (e.g., Y2O3:Yb3+/Er3+ and Gd2O3:Yb3+/Er3+).74,75 However, the size of the oxysulfide or oxide particles was at the submicron level, which limited applications. The necessary requirements for material selection in practical bioimaging are small size, bright luminescence, and biological safety. Recently, with the rapid development of synthesis techniques, fluoride-based UCNPs with smaller size but high-quality luminescence have been explored extensively and widely used in cell, tissue, and animal imaging. In contrast to oxysulfides or oxides, fluorides are considered better host materials for the doping of lanthanide ions to achieve intense UC emissions, owning to their low phonon energies and the resulting minimization of quenching of the excited state of the lanthanide ions. One of the first reports of in vivo imaging of UCNPs in small animals reported spot measurements of UCNPs injected subcutaneously in rats by Zhang et al. in 2008, which showed much deeper penetration under 980 nm excitation compared to commercial green-emitting QDs excited by UV.76 Although the NIR excitation light for UC materials has strong penetration ability, the UV/VIS UC emissions are still easily absorbed by biological samples, which definitely limits their further applications in the observation of deep biological tissues. For small-animal in vivo imaging, highly sensitive NIR–NIR systems have attracted increasing attention, because both the excitation and emission light is located in the NIR region. For this purpose, Tm3+ ions are frequently chosen as dopants since they can exhibit strong NIR emission between 750 and 850 nm due to the transition from 3H4 to 3H6 under CW laser excitation at 980 nm. Nyk et al. reported the use of 20–30 nm NaYF4 NPs doped with Tm3+ and Yb3+, which has an emission around 800 nm for both in vitro and in vivo imaging (Figure 1.2).77 High-contrast photoluminescence (PL) imaging was possible in cells and small animals due to the better tissue penetration properties achieved since both the excitation and emission are in the NIR region. After that, in vivo whole-body imaging of small animals has been successfully realized based on Tm3+-doped NaYF4,78 NaGdF4,79 NaLuF4,80 and NaYbF4 NPs.81
Besides 750–850 nm emission from Tm3+, the red emission (625–690 nm, centered at 660 nm) from Er3+ or Ho3+ is also ideal for bioimaging. Therefore, much effort has also been devoted to enhancing the red emission of Er3+ or Ho3+—in other words, to achieving an enhanced red/green (R/G) ratio in the Yb/Er codoped upconversion system. In 2011, Liu et al.82 described an oil-based synthetic method for the preparation of KMnF3 nanocrystals with lanthanide dopants homogeneously incorporated in the host lattice. With Yb3+/Er3+ doping, blue and green emissions of Er3+ disappeared completely, suggesting an extremely efficient exchange-energy transfer process between the Er3+ and Mn2+ ions, which can be largely attributed to the close proximity and effective mixing of wave functions of the Er3+ and Mn2+ ions in the crystal host lattices (Figure 1.3a). Besides Yb3+/Er3+, Yb3+/Ho3+ and Yb3+/Tm3+ doped KMnF3 nanocrystals were synthesized, respectively. Importantly, these nanocrystals also displayed single-band emissions involving the 5F5 → 5I8 (centered at 650 nm) transition in Ho3+ (Figure 1.3b) and the 3H4 → 3H6 (centered at 800 nm) transition in Tm3+ (Figure 1.3c). As a result of efficient energy transfer between the dopant ion and host Mn2+ ion, remarkably pure single-band upconversion emissions were generated in the red and NIR spectral regions. The complete lack of short-wavelength emission of these lanthanide-doped nanocrystals in the visible spectral region provides a platform for promising applications in biolabeling studies, for which imaging at different sample depths is required. One year later, Zhao et al.83 reported red-emission UCNPs based on NaYF4 system, which is considered as one of the best host matrices for the upconversion process. In this work, Mn2+ served as a dopant to influence the growth dynamics of the crystalline phase and size of the resulting UCNPs, rather than host material. The Mn2+-doped NaYF4:Yb3+/Er3+ (18/2 mol%) NPs were obtained by using a modified liquid–solid solution (LSS) solvothermal strategy: with the increased doping amount of Mn2+ ions, the phase transformation from hexagonal to cubic was obvious. Pure cubic NaYF4 was obtained when the level of Mn2+ ions reached 5 mol% and no obvious extra diffraction peaks were detected even when the Mn2+ ion concentration increased to 30 mol%, indicating the formation of a Y–Mn solid solution. Interestingly, the R/G ratio gradually increased from 0.83 to 163.78 with increasing Mn2+ dopant content. The appearance of single-band red upconversion emission suggests that the exchange-energy transfer process between the Er3+ and Mn2+ ions is extremely efficient, which agrees with the conclusion from Liu et al. These red-emitting UCNPs can penetrate deeper than 10 mm when used for imaging in vivo, which is rarely reported in other papers (Figure 1.3d–f). Later in 2014, Hao et al.84 extended these Mn2+-doped UCNPs to other host materials, such as NaLuF4 and NaYbF4. They also observed large enhancements in overall UC luminescent spectra of Mn2+-doped UCNPs (∼59.1 times for the NaLuF4 host, ∼39.3 times for the NaYbF4 host compared to the UCNPs without Mn2+ doping), mainly due to remarkably enhanced luminescence in the red band. Although great advances have been made in these three papers,82–84 simultaneous control of the structure (nanocrystal size, shape, and phase) and enhancement in upconversion luminescence especially dominated by red emission in UCNPs with a fixed formula is still a great challenge. In 2015, Tian and his coworkers successfully synthesized a novel kind of small hexagonal-phase Mn2+-doped NaYbF4:Er3+ UCNPs with bright and red emission by a modified codeposition method.85 This method was more controllable and convenient than the hydrothermal or solvothermal methods used previously82–84 (Figure 1.3g–l). Moreover, Tian et al. found that the dopant Mn2+ ions have a negligible effect on the phase structure since all the diffraction peaks of the samples still correspond to the pure hexagonal phase without admixture of cubic phase or other impurities. This finding was totally different from Zhao's and Hao's results, and it is well known that hexagonal-phase materials always exhibit higher upconversion efficiency relative to their cubic-phase counterparts. Therefore, Tian et al. concluded these hexagonal-phase Mn2+ doped NaYbF4:Er3+ UCNPs may highly desirable in biomedicine, especially in bioimaging. Most recently, Rai et al.86 reported significant enhancement in the red upconversion emission of Er3+ in NaSc0.8Er0.02Yb0.18F4 UCNPs through resonance energy transfer and plasmonic effect from Au NPs. Attachment of Au NPs on the surface of UCNPs gave two advantages: reduction in green band (through resonance energy transfer with efficiency 31.54%) and enhancement in red band (through the plasmonic effect). It gave a R/G ratio of nearly 20 : 1 (almost single-band red UC), which is quite promising for imaging applications.
1.2.2 Single-Band UCNPs
Single-band UCNPs have only one emission band under NIR excitation. Triply ionized lanthanide ions in UCNPs typically have emission line widths of 10–20 nm (FWHM) in the visible portion of the spectrum, which is approximately half the line width observed for QDs (25–40 nm) and much narrower than the line width observed for organic dyes (30–50 nm). This feature increases the number of distinguishable emission bands within a specific spectral bandwidth, enabling a large number of multiplexed detections. Although UCNPs have shown significant advantages over the traditional organic fluorophores or QD fluorescent biolabels, a problem remains: each lanthanide ion has a unique set of energy levels and generally exhibits a set of sharp emission peaks with distinguishable spectroscopic fingerprints. To minimize this spectral interference, in 2015 our group reported a general and simple method of achieving single-band upconversion emission with different colors by coating the upconversion nanocrystals with a screen layer containing an organic dye with a high molar absorption coefficient as a nanofilter to remove the unwanted emission bands.87 As a result of the efficient reabsorption of the organic dye, remarkably pure single-band upconversion emissions can be generated in the blue, green, and red regions. The organic dyes were selected on the criteria of overlapping absorption spectra with only one of the dual emission bands of the nanocrystals, with a high molar absorption coefficient (in the range of 105 M−1 cm−1). A pure silica spacer layer was used to prevent Förster resonance energy transfer (FRET) between the filtered upconversion emission band and the fluorescent dye-doped screen layer. To obtain green single-band emission, nickel(ii) phthalocyanine-tetrasulfonic acid tetrasodium salt (NPTAT) organic dyes with a maximum absorption wavelength (λmax) of 657 nm were added with tetraethyl silicate (TEOS) to form NPTAT-doped silica nanofilters on the β-NaGdF4:20% Yb, 2% Er@NaGdF4@SiO2 NPs to filter the red emission band efficiently. With the β-NaGdF42:0% Yb, 2% Er@NaGdF4@SiO2@NPTAT-doped SiO2 nanostructure, only the narrow green emission centered at 540 nm was observed, in stark contrast to the dual upconversion emission bands of β-NaGdF4:20% Yb, 2% Er@NaGdF4. To obtain the blue and red emission single-band UCNPs, β-NaGdF42:0% Yb, 0.2% Tm@NaGdF4, and α-NaYbF41:0% Er@NaYF4 nanocrystals with strong blue to red and red-to-green upconversion emission ratios were first prepared. After coating the pure SiO2 layers, nanofilters doped with NPTAT and rhodamine B isothiocyanate were used to filter the red and green emissions to obtain the final blue and red single-band UCNPs, respectively. This general approach permits not only the removal of minor emission peaks away from the main peaks using appropriate nanofilters, but also the alternative removal of the main peaks to leave the minor peaks for single-band upconversion emission. For example, green emission single-band UCNPs can be obtained by coating the α-NaYbF41:0% Er@NaYF4 nanocrystals with NPTAT-doped nanofilters. Besides green (550 nm) emission single-band UCNPs, remarkably pure single-band upconversion emissions can also be generated in the blue (480 nm) and red (650 nm) regions.
Significantly, in this work, we have demonstrated the use of single-band UCNPs for the multiplexed detection of three tumor biomarkers in both cultured human breast cancer cells and paraffin-embedded clinical tissue sections. The simultaneous quantification of estrogen receptor (ER), progesterone receptor (PR), and HER2 receptor expression levels in the breast cancer cell specimens correlated closely with the results of the traditional western blot method. Furthermore, the application of conjugated single-band UCNPs and quantitative spectroscopy may be more accurate than immunoenzyme-based immunohistochemical (IHC) methods for the simultaneous quantification of proteins present at low levels in cancer cells and tissue specimens. Thus, single-band UCNP-based technology may be well suited to the molecular profiling of tumor biomarkers in vitro and represent a clinically translational application of upconversion nanomaterials for cancer prognosis. The ability to detect multiple target proteins in small samples of cancer tissues could enable more effective therapeutic decisions when used in combination with regular IHC methods. The next step is to conduct large-scale clinical studies to establish protocols and practices for single-band UCNP-based molecular pathology.
In the work discussed above, single-band UCNPs were used for multispectral in vitro biodetection, but we believe that single-band UCNPs may also hold great promise in multispectral bioimaging which requires no overlapping signals between different imaging agents (Figure 1.4).
1.2.3 UCNPs Excited at Another Wavelength Range
Traditionally, highly efficient UCNPs require a NIR laser at a wavelength of about 980 nm as the excitation source since the sensitizer ion (Yb3+) has a high absorption cross-section in its absorption band. Unfortunately the 980 nm laser light is strongly absorbed by water and biological specimens. Thus, 980 nm excitation has associated problems such as limited penetration depth and tissue damage due to sample overheating. Therefore, excitation band tuning of UCNPs into an appropriate range is also important for improving their performance. Zou et al. reported the concept of a UCNP where an organic NIR dye is used as an antenna to harvest the NIR photons within a broad band (740–850 nm) for the β-NaYF4:Yb,Er NPs in which the upconversion occurs (Figure 1.5a). The overall upconversion by the dye-sensitized NPs is dramatically enhanced (by a factor of ∼3300) as a result of increased absorptivity and overall broadening of the absorption spectrum of the upconverter.88 Prasad's group also reported the intense upconversion PL in colloidal LiYF4:Er3+ nanocrystals under excitation at 1490 nm telecom wavelength (Figure 1.5b). The intensities of two- and three-photon anti-Stokes upconversion PL bands are higher than or comparable to that of the Stokes emission under excitation with low power density in the range 5–120 W cm−2. The QY of the upconversion PL was measured to be as high as ∼1.2 ± 0.1%, which is almost four times higher than the highest upconversion PL efficiency (0.3 ± 0.1%) reported to date for lanthanide-doped nanocrystals in 100 nm hexagonal NaYF4:Yb3+,Er3+ using excitation at 980 nm.89 Later in 2015, the same group reported a novel multilayer core–shell design to broadly upconvert infrared light at many discrete wavelengths into visible or NIR emissions. It utilized hexagonal-phase core/multishell NaYF4:10% Er3+@NaYF4@NaYF4:10% Ho3+@NaYF4@NaYF4:1% Tm3+@NaYF4 NPs. These core–multishell NPs can emit UC PL emission from Ho3+ (5F5 → 5I8, 625–685 nm range), Tm3+ (3H4 → 3H6, 760–860 nm range), and Er3+ (4S3/2 → 4I15/2, 510–570 nm range) when excited at ∼1120–1190 nm (due to Ho3+), ∼1190–1260 nm (due to Tm3+), and ∼1450–1580 nm (due to Er3+), respectively. The excitation light could collectively cover a broad spectral range of about 270 nm in the NIR range.90 Zhan et al. also demonstrated a new and promising excitation approach for better NIR-to-NIR UC PL in vitro or in vivo imaging employing a cost-effective 915 nm laser. This novel laser excitation method led to much less heating of the biological specimen and greater imaging depth in the animals or tissues because water absorption was quite low (Figure 1.5c).91
1.2.4 Nd3+ Sensitized UCNPs
Recently, researchers have paid more attention to the 800 nm excitation UCNPs. Notably, Nd3+ has multiple NIR excitation bands at wavelengths shorter than 980 nm, such as 730, 808, and 865 nm, corresponding to transitions from 4I9/2 to 4F7/2, 4F5/2, and 4F3/2, respectively. At all of these wavelengths, water absorption is lower, and the typical absorption coefficient is 0.02 cm−1 at 808 nm, in contrast to 0.48 cm−1 at 980 nm.92 Consequently, the laser-induced heating effect, especially for biological tissues, is expected to be greatly minimized. Meanwhile, Nd3+ has a large absorption cross-section in the NIR region (1.2 × 10−19 cm2 at 808 nm) compared to that of Yb3+ (1.2 × 10−20 cm2 at 980 nm), which also benefits the efficiency of the Nd3+-sensitized upconversion process.
Han et al. found when Nd3+, Yb3+, Er3+/Tm3+ were codoped in NaYF4, upconversion distinguishable by the naked eye can be observed upon the 800 nm excitation (Nd3+ has a broad excitation wavelength in the range of 790–810 nm). In these novel UCNPs, Nd3+ serves as an 800 nm photon sensitizer and Yb3+ as a bridging ion, with the energy transfer procedure performing as Nd3+ → Yb3+ → Er3+/Tm3+. However, the doping limit of the Nd3+ is only 1% because of concentration quenching.93 Apparently, novel structure needs to be designed to increase the doping concentration of Nd3+. In 2013, Liu et al. reported a new type of core–shell UCNPs (Figure 1.6a). Through spatially confined doping of Nd3+, which can be effectively excited at 795 nm, they claimed that the active NaYF4:Nd3+ shell layer can effectively prevent surface quenching of Yb3+ emission and can simultaneously promote the transfer of excitation energy to Yb3+ ions, which significantly enhances the upconversion; also, 800 nm is better than 980 nm for heat generation, as confirmed by cell irradiation experiment (Figure 1.6b and c).94 Almost at the same time, Yan and his co-workers reported a similar core–shell structure (Figure 1.6d) with a similar design idea by doping Nd3+ in the shell to ensure successive Nd3+ → Yb3+ → activator energy transfer. In vivo imaging of a nude mouse subcutaneously injected with UCNPs showed that comparable photon numbers can be measured when irradiated with 980 nm laser and 808 nm laser, respectively (Figure 1.6e). In addition to the upconversion process, these authors also performed downconversion PL of lanthanide-doped NPs. In vivo NIR imaging of a nude mouse injected with UCNPs showed high signal-to-noise ratio under 808 nm laser excitation (Figure 1.6f–h).95
These results indicate that excitation at 800 nm is indeed a good future direction for development of UCNPs in biomedical imaging. However, more theoretical research has shown that the efficiency limitation of Nd3+-sensitized UCNPs is the “energy back-transfer” phenomenon, which can efficiently transfer energy from activators back to 4IJ manifolds of Nd3+, such as from Er3+ to Nd3+. As a result, the doping concentration of Nd3+ in UCNPs must be constrained to a very low level (<1%) to minimize the quenching of the excitation energy. Therefore, directly doping Nd into UCNPs always results in much lower upconversion luminescence under 800 nm NIR laser irradiation than that of the conventional 980 nm-excited UCNPs in the same conditions.
In 2014, a breakthrough was made by Zhao's group. They first developed a well-defined NaYF4:Yb,X@NaYF4:Yb@NaNdF4:Yb (X = Er3+, Tm3+, Ho3+) core–shell–shell structure (Figure 1.7a and b) to separate the activator (Er3+, Tm3+, Ho3+) and sensitizer (Nd3+) into different layers, which enables efficient harvesting of NIR light, suppresses the cross-relaxation between the sensitizer and activator, and finally results in the generation of efficient upconversion emissions under 800 nm CW laser excitation (Figure 1.7c).96 They found that the emission intensity of NaYF4:Yb,Er@NaYF4:Yb@NaNdF4:Yb core–shell–shell (ErCSS) NPs (∼16 nm) with optimized shell thickness shows enhancement factors of ∼2000 (compared with the conventional NaYF4:Yb,Er@NaYF4 core–shell (ErCS) NPs), ∼100 (compared with the Nd/Yb/Er tri-doped UCNPs) and ∼8 (compared with the Nd-coated core–shell UCNPs) under 800 nm irradiation. Notably, the upconversion emission of the 800 nm-excited ErCSS NPs (∼16 nm) is ∼7 times higher than that of the conventional 980 nm-excited ErCS NPs (∼20 nm, hexagonal phase) at a low excitation power (0.1 W). Moreover, other lanthanide ions conventionally used for generating upconversion emissions, including Tm3+ and Ho3+, can also serve as the activator in an Nd3+-sensitized core–shell–shell system (Tm: TmCSS; Ho: HoCSS), because the efficient energy transfer from Yb3+ to these lanthanide ions can still be facilitated. Similar enhanced upconversion emission could be observed in Tm (∼4.6 times) and Ho (∼2 times) doped core–shell–shell nanocrystals. An explanation of the high efficiency is that the Nd3+ and the activators (Er3+, Tm3+, Ho3+) were separated by a core–shell structure, in which the Nd3+ was confined in the shell and the activators were embedded in the core. Therefore, the non-radiative processes resulting from Er–Nd interactions are largely impeded. These characteristics lead the authors to believe that these 800 nm-excited core–shell–shell NPs with improved optical performance will outperform conventional 980 nm-excited UCNPs and play an important role in the development of fluorescent probes for future bioimaging applications.
Recent advances in Nd3+-sensitized UCNPs are 800 nm-excited UV/blue and red light, which are considered good light sources for stimulation–response systems (such as controlled drug delivery or photodynamic therapy) and bioimaging, respectively. In 2013, Wang et al. reported 808 nm-excited UV/blue emission from a novel NaYbF4:50% Nd@Na(Yb,Gd)F4:Tm3+@NaGdF4 core–shell–shell nanoparticle.98 They claimed that the appearance of upconversion emission peaks in the UV spectral region is largely owing to the core–shell structure, which suppresses deleterious cross-relaxations in the NPs. When the Tm3+ was homogenously doped with Yb3+ and Nd3+ ions in the core layer, UV upconversion emission peaks essentially disappeared. Strikingly, for the first time they realized UV emission of Tm3+ at around 300 nm in NPs by 808 nm diode laser excitation through careful control of the doping concentration of Tm3+. In addition, the establishment of a population at a high-lying energy state of Tm3+ also further enables an energy cascade in the Gd sublattice followed by energy trapping and optical emission from common lanthanide ions including Tb3+, Eu3+, and Dy3+. Later in 2015, they applied their core–shell–shell nanostructure in 800 nm-excited red UCNPs.99 They have assessed a series of NaYbF4:Nd@NaGdF4:Yb/Er@NaGdF4 NPs with varying Yb3+ concentrations in the inner shell layer. The red emission of Er3+ gradually dominated the spectra with increasing Yb3+ concentration from 18 to 78 mol%, which corresponds to intensity ratios of red-to-green emission from 0.5 to 0.7, 0.85, and 1.9. The steady increase in the red/green ratio was mainly induced by the 4S3/2 + 2F7/2 → 4I13/2 + 2F5/2 and 4I13/2 + 2F5/2 → 4F9/2+ 2F7/2 cross-relaxations at elevated Yb3+ concentrations. Further increasing Yb3+ concentration did not lead to noticeable improvement in red emission of Er3+, probably due to the poor shell quality as a result of the fast shell deposition process in the absence of Gd3+ cofactors. They also examined relevant NPs comprising high concentrations of Er3+ in the inner shell layer, which are known to induce 4S3/2 + 4I9/2 → 4F9/2 + 4F9/2 cross-relaxation between Er3+ ions for promoting the red emission of Er3+. PL investigation showed that the emission spectra of Er3+ can only be marginally tuned by varying the Er3+ concentration, accompanied by a decrease in the overall emission intensity at high Er3+ content. Taken together, the optimal Yb/Er concentration was determined to be 78/2 mol%. In their in vitro experiment, they found the optical emission can be clearly observed when a 5 mm sample of pork muscle tissue is placed between the cells and the irradiating laser. The upconversion emission signals were still detectable after the thickness of the muscle tissue was increased to 10 mm. As a control experiment, classical NaYF4:Yb/Er (18/2 mol%)@NaYF4 core–shell NPs can hardly be detected when the laser beam is blocked by a muscle tissue as thin as 5 mm. Almost at the same time, Chen et al. presented another strategy to achieve 808 nm-excited single-band red upconversion luminescence of Ho3+ via Ce3+ to change the red/green ratio in the NaGdF4:Yb/Ho/Ce@Yb/Nd:NaYF4 active-core@active-shell nanoarchitecture (Figure 1.7d).97 The doping of Ce3+ plays a key role in the realization of Ho3+ single-band red luminescence via the efficient cross-relaxation processes between Ce3+ and Ho3+: that is, Ho3+5:S2/5F4 + Ce3+2:F5/2 → Ho3+5:F5 + Ce3+2:F7/2 and Ho3+5:I6 + Ce3+2:F5/2 → Ho3+5:I7 + Ce3+2:F7/2. This nanoarchitecture enables the spatial separation between Ho3+ and Nd3+ and subsequently the high-content Nd3+ doping to efficiently improve upconversion luminescence, which might finally provide highly attractive luminescent biomarkers for bioimaging without the problematic overheating effect.
1.3 Lanthanide Downconversion Nanoparticles (DCNPs)
Lanthanide-based NIR downconversion nanoparticles (DCNPs) emerged a little later than UCNPs. The synthesis and surface modification procedures successfully used for UCNPs can also be well utilized on DCNPs. Ions of almost all the 15 lanthanide possess the downconversion property, and their emission wavelength can be located in the UV, visible, NIR I, NIR II and even mid-IR range.100–102 Typically, 865–900 nm, 1060 nm, and 1300 nm from Nd3+ (4F3/2 → 4I9/2, 4F3/2 → 4I13/2, and 4F3/2 → 4I15/2, respectively), 900–1000 nm from Yb3+ (2F5/2 → 2F7/2), 1185 nm from Ho3+ (5I6 → 5I8), 1310 nm from Pr3+ (1G4 → 3H5), 1475 nm from Tm3+ (3H4 → 3F4), and 1525 nm from Er3+ (4I13/2 → 4I15/2) are common NIR emission wavelengths in the spectral database of the lanthanide ions. However, the luminescence efficiency of these wavelengths varies greatly, and developing efficient DCNPs with proper emission wavelengths is an important issue in this field.
The first attempts to utilize DCNPs as biomedical imaging agents (NIR I or NIR II) have been made in the last decade. As early as 2002, Veggel et al. reported Nd3+-doped LaF3 NPs with both NIR I and NIR II emission under 514 nm laser excitation, as well as Er3+- or Ho3+-doped NPs. Because the interesting luminescence emission is in the telecommunication window (i.e., Er3+ at 1530 nm, Nd3+ at 1330 nm, and Ho3+ at 966 nm and 1450 nm), they had considered them as promising materials for polymer-based optical components.103 In 2006, inspired by Veggel's work, Wang et al. developed a simple method to synthesize LaF3:Nd3+ in aqueous solution at low temperature, and have pointed out that this kind of DCNP would have potential application in biomedical imaging. The emission of Nd3+-doped LaF3 nanocrystals is located in NIR II under 802 nm laser excitation.104 Later in 2008, NdF3 NPs were synthesized in aqueous solution by a similar method as used to synthesize traditional Nd3+-doped LaF3 NPs. NdF3 NPs showed no doping concentration quenching effect; moreover, the vibrational quenching caused by the O–H groups on the surfaces of the NdF3 NPs can be suppressed after coating with silica shells. For deep-tissue imaging, mice were injected intramuscularly and intraperitoneally with 100 µL of NdF3/SiO2 NPs (1.0 µg mL−1) into the thigh and abdominal cavity respectively. NIR signals (1050 nm) from the deep tissues of the thigh and abdominal cavity can both be clearly distinguished from the tissue autofluorescence under 730 nm excitation.105 Later in 2013, Nd3+-doped DCNPs based on a GdF3 host matrix were realized by Mimun et al.106 The GdF3:Nd3+ NPs were small, with an average size of 5 nm, and formed stable colloids that lasted for several weeks without settling, enabling their use for several biomedical and photonic applications. Their excellent NIR properties, such as a nearly 11% QY the 1064 nm emission, make them ideal contrast agents and biomarkers for in vitro and in vivo NIR optical bioimaging. The nanophosphors, which were coated with poly(maleic anhydride-alt-1-octadicene) (PMAO), were implemented in cellular imaging, showing no significant cellular toxicity for concentrations up to 200 mg mL−1. A proof-of-concept experiment for imaging through tissue was conducted by placing the GdF3:Nd3+ NPs under varying thicknesses of pig skin, ranging from 0.67 to 5 mm. Emission spectra were collected through each thickness, and the 1064 nm emission was easily discernible even at the greatest tissue thickness of 5 mm. Furthermore, the incorporation of Gd into the nanocrystalline structure endowed these NPs with exceptional magnetic properties, making them ideal for use as magnetic resonance imaging (MRI) contrast agents. Almost at the same time, the same group reported their investigation of the downconversion absolute quantum yields measurement on the powder, PMAO-coated powder, and colloidal solution states of GdF3:Nd3+ NPs.107 The maximum total absolute downconversion QY of 10.2 ± 1.5% was measured for the GdF3:Nd3+ nanophosphor powder at an excitation power density of 12.74 ± 2.0 W cm−2 at 800 nm excitation. Similarly, downconversion QYs of 5.02 ± 0.75% and 2.2 ± 0.33% were measured at an excitation power density of 5.3 ± 0.8 and 1.4 ± 0.2 W cm−2, respectively. With the known QY 10% for IR-140 dye in the spectral range of 862–1013 nm at an excitation power of 150 mW under 800 nm excitation, a comparison method was also implemented to check the accuracy of the measurement. Comparison measurement for GdF3:1% Nd3+ powder shows that the downconversion QY in GdF3:1% Nd3+ is 5.8 ± 0.87% at 150 mW (4.77 W cm−2) excitation under 800 nm, which is very close to the measured QY at 5.3 W cm−2 using integrating spheres for GdF3:1% Nd3+ powder. Scaling of the downconversion emission spectra revealed that the 1064 nm emission from Nd3+ represents around 90% of the overall downconversion emission intensity. In addition, compared with the upconversion QY of 0.005 ± 0.0005% (at 150 W cm−2) reported by van Veggel et al. for β-NaYF4:20% Yb3+/2% Er3+ of 8–10 nm sized particles, the downconversion QY for GdF3:1% Nd3+ nanophosphor powder is 2000 times higher even at an excitation power density of 12.74 ± 2.0 W cm−2 at 800 nm excitation.108 This shows that these particles have a higher QY within the biological window which yield more photon counts (information density) for bioimaging applications compared to UCNPs. Furthermore, the comparison method was implemented to measure the downconversion QY for colloidal GdF3:1% Nd3+ with respect to the reported QY for the dye IR-140 to mimic the experimental conditions. Using the QY of 10% reported for the IR-140 at 150 mW (4.77 W cm−2) excitation under 800 nm, the downconversion QY for GdF3:1% Nd3+ at a concentration of 0.05 mg mL−1 was measured to be 1 ± 0.05%. Similarly, the QY of 1.5 ± 0.075% was measured for GdF3:1% Nd3+ at 255 mW (8.28 W cm−2). This verifies that the QY for colloidal GdF3:1% Nd3+ is dependent on concentration and excitation power density, indicating that DCNPs yield more photon counts (information density) for bioimaging applications than UCNPs. The authors also have measured the NIR emission spectra obtained with and without the additional PMAO coating for the GdF3:1% Nd3+ at an excitation power density of 12.74 ± 2.0 W cm−2. Coating the NPs with PMAO does not significantly change the measured downconversion QY, which is important since the polymer coating is essential for making the particles biocompatible.
Although considerable research has been done on LaF3:Nd3+, GdF3:Nd3+, and NdF3, morphology control of the NPs is still a big problem for this type of materials. Thanks to the well-developed synthesis methods for UCNPs, uniform and monodispersed NaReF4:Nd3+ NPs (where Re = rare earth elements) have been obtained, and have attracted increasing attention in recent years.109–111 In 2012, Prasad et al. reported highly efficient NaGdF4:Nd3+@NaGdF4 for NIR–NIR biomedical imaging.112 Unlike LaF3:Nd3+, these novel DCNPs benefit from controllable morphology and even ∼3 nm uniform NaGdF4 shells have been successfully synthesized (Figure 1.8a). These DCNPs exhibited spectrally sharp, photostable, and large Stokes-shifted NIR PL at 900, 1050, and 1300 nm when excited at 740 nm (Figure 1.8b). The absolute QY of this NIR-to-NIR downconversion PL was evaluated to be as high as 40% for core–shell NaGdF4:Nd3+@NaGdF4 NPs dispersed in hexane and 20% for ligand-free NaGdF4:Nd3+@NaGdF4 NPs dispersed in water. The high luminescent efficiency in NPs was realized by effective suppression of non-radiative losses originating from surface passivation and cross-relaxation between Nd3+ dopants, as revealed by the PL steady-state and time-resolved studies. A facile high-contrast NIR-to-NIR imaging of HeLa cells and a nude mouse were demonstrated by the authors, utilizing excitation from an incoherent light source through observation of NIR PL at 900 nm (Figure 1.8c–e).
Recently, some attempts have been made to explore the downconversion optical property of the traditional UCNPs. Nagasaki et al. had reported the application of Y2O3:Yb3+,Er3+ as a UCNP biomedical imaging nanoprobe in 2008.113 In 2011, they used the same kind of NPs for the downconversion bioimaging research. A strong NIR signal (1550 nm) can be observed 24 h after intravenous injection of DCNPs upon the irradiation of 980 nm laser.114 Although this work is at an early stage, it also indicates that other lanthanide ions, such as Er3+, may hold promise for NIR bioimaging. Er3+ possesses a stable energy level in the NIR range (4I13/2), which can emit NIR light in the wavelength range of 1450–1650 nm. In 2014, exciting work on the downconversion phenomenon of UCNPs was published by Moghe and his colleagues. They first developed a library of rare earth nanomaterials with tunable, discrete SWIR (short-wavelength infrared, 1000–2300 nm, including NIR II) emissions and proceeded to evaluate their optical performance for several clinical imaging applications including real-time, multispectral in vivo SWIR imaging.115 The rare earth nanomaterials they used were NaYF4:Yb3+, Ln3+ (Ln = Er, Ho, Tm, or Pr)@NaYF4 core–shell NPs. By doping the NaYF4 core with Yb and one of several other rare earth elements, such as Er, Ho, Tm, or Pr, the emission properties of rare earth nanomaterials can be tailored in both the SWIR and visible ranges. The fluorescence of rare earth nanomaterials occurs following the resonant transfer of excitation energy from a sensitizer (Yb) to an activator dopant such as Er, Ho, Tm, or Pr. The relaxation from an excited state results in the generation of SWIR emissions that are unique to the specific rare earth activator. Hexagonal-phase NaYF4:Yb3+,Er3+ NPs were among the brightest SWIR-emitting rare-earth-doped phosphors (with an optical efficiency >1.1%), and therefore were chosen to illustrate not only the benefits of SWIR compared to conventional optical imaging methods but also the biomedical potential for SWIR-based imaging approaches. Compared to other SWIR emitters, the rare earth nanomaterials presented in this work are considerably more effective at generating SWIR emissions than single-walled carbon nanotubes (SWNTs) and IR-26, an organic SWIR dye. Only highly toxic lead-based QDs matched the SWIR emission power output of the rare earth nanomaterials. This group further investigated the attenuation properties of actual biological tissues in the SWIR by measuring the absorbance spectra of excised tissue obtained from a mouse exhibiting pigmented tumor lesions. The majority of tissue samples exhibit markedly low attenuation at 1000–1350 nm as well as at 1500–1650 nm, effectively extending the wavelength region of lowered attenuation within the second ‘tissue-transparent window’ of SWIR. Furthermore, strong absorbers in the tissue, such as melanin in tumor samples and hemoglobin in blood samples, exhibited strong attenuation in the visible range whereas attenuation in the SWIR was weak. Importantly, <0.4% of NIR I light penetrated through 0.5 cm of pigmented tumor tissue compared to ∼80% transmittance achieved by SWIR, suggesting that NIR I has limited use for detecting optical probes in melanin-containing tissues. Furthermore, they determined the actual tissue penetration depth of SWIR compared to the current standard imaging wavelength region by measuring the intensity of SWIR and NIR I light through tissue phantoms composed of both scattering and absorbing agents. The intensity of the NIR I signal rapidly diminishes over increasing phantom depth, with complete signal loss occurring by 5 mm. In contrast, the high SWIR emission from the rare earth nanomaterial pellet saturates the camera's detector for phantom tissues at 5 mm or less. Notably, SWIR signal was detectable through 1 cm of phantom tissue, whereas no signal above background was seen using NIR I. In the in vivo SWIR imaging experiment, rare earth nanomaterials also showed good performance. Using a series of image-processing algorithms, SWIR signal intensities as a function of concentration of rare earth nanomaterials were found to exhibit a linear relationship in both tissue phantoms and subcutaneously injected mice, with a detection threshold at ∼3 nM rare earth nanomaterials at an excitation power density of 0.14 W cm−2 and camera exposure time of ∼50 ms per frame. In comparison to QDs and SWNTs with reported detection limits of ∼5 nM and 6 nM respectively, rare earth nanomaterials can be detected at a lower concentration under comparable excitation. After injection, SWIR emissions were first identified in the tail vein (5 s) before clearing the vasculature to enter the heart and lungs (10 s). The beating of the heart in the chest of the mouse was visualized by pulsing SWIR emissions captured in real time (Figure 1.9). Over the course of 60 s, the SWIR signal became progressively more intense in organs such as the liver and spleen, which are part of the reticuloendothelial system that mediates nanoparticles. In contrast, the accompanying visible signal from UCNPs was notably absent, probably due to absorption and scattering losses caused by blood and tissue components.
Although great advances have been made by Moghe and his colleagues, 980 nm-excited lanthanide-based NIR nanomaterials, both UCNPs and DCNPs, suffer from the same problem: the water in biological structures would overwhelmingly attenuate 980 nm light, and transform its energy into local heat which could damage cells and tissues. For efficient bioimaging, it is therefore essential to develop a novel lanthanide-based SWIR probe with excitation source optimization. For UCNPs, a solution has already been mentioned above: Nd3+-sensitized UCNPs. The same idea can also be used for DCNPs. In 2014, our group reported a novel kind of β-NaGdF4/Na(Gd,Yb)F4:Er/NaYF4:Yb/NaNdF4:Yb C/S1/S2/S3 DCNPs as an efficient 800 nm NIR to 1525 nm SWIR probe for in vivo bioimaging.116 C/S1/S2/S3 DCNPs were synthesized using the epitaxial seeded growth method, and are composed of the NaGdF4 core (seed for epitaxial growth), a Na(Gd,Yb)F4:Er shell (S1, the SWIR-emitting layer), a NaYF4:Yb shell (S2, the energy transfer layer), and a NaNdF4:Yb shell (S3, the energy absorption layer) (Figure 1.10). After absorbing 800 nm excitation energy (Nd3+, 4I9/2 → 4F5/2), the S3 transfers its energy into the inner S2 layer (Nd3+ → Yb3+, 2F7/2 → 2F5/2). Energy is transferred within this layer by the codoped Yb3+ until the Er3+ in the S1 is sensitized (Yb3+ → Er3+, 4I15/2 → 4I11/2). This results in relaxation from the excited state of Er3+ via the release of a 1525 nm (4I13/2 → 4I15/2) photon along with phonon vibration.
As the 1525 nm SWIR PL occurs following the resonant transfer of excitation energy from the Yb sensitizer to the Er activator, efficient resonant energy transfer theoretically requires a high concentration of the Yb sensitizer. Therefore, in order to realize the highly efficient 1525 nm SWIR PL, we needed to find an optimal doping concentration for the Yb sensitizer. In our case, the doping ratio of Yb3+ and Er3+ can be tuned easily without changing the morphology due to the epitaxial seeded growth method. If we had instead used the Na(Gd,Yb)F4:Er as the starting core of this core–shell nanostructure, the particle size of the Na(Gd,Yb)F4:Er would have been uncontrollable when the Yb : Er doping ratio was changed, resulting in non-comparable spectral data because the PL properties depend on the size of the lanthanide-based NPs. With delicate adjustment of Yb : Er doping ratio, the best composition of the S1 layer was optimized to NaYbF4:2Er to realize the most efficient SWIR emission. Similar experiments were also carried out to optimize the Yb and Nd dopant concentrations in the S2 and S3 layers. We finally found that the optimum composition for efficient 1525 nm SWIR emission under 800 nm excitation is NaGdF4/NaYbF4:2Er/NaYF4:10Yb/NaNdF4:10Yb.
We next studied the penetration depth of our C/S1/S2/S3 NPs. A pellet of the NPs was excited at 800 nm and the 1525 nm signal was imaged using an InGaAs camera through increasing thickness of tissue (pork slices). Notably, SWIR signals from C/S1/S2/S3 NPs were clearly detectable through 1.8 cm of pork tissue, much deeper than the 1.0 cm reported for the 1525 SWIR probe excited by a 980 nm laser under comparable excitation power density (∼30 µW), further confirming that the 800 nm excitation source is more suitable for in vivo applications. As mentioned above, Moghe et al. demonstrated that 1525 nm SWIR transmits more effectively through tissue phantoms than 800 nm NIR light with the same intensity. That was the first experimental demonstration of the imaging advantages of SWIR due to the reduced tissue absorbance and scattering within this second window compared with that of an emitter in the NIR I window. However, there are still no experimental results to evaluate the effect of wavelength on the bioimaging performance around the second NIR window. Here, we systematically evaluated the dependence of in vitro and in vivo bioimaging performance on the SWIR emission wavelength in comparison to the previously reported NaGdF4:Nd/NaGdF4 NPs with 1060 nm SWIR signal, also upon 800 nm excitation (Figure 1.11a and b). For the 1060 nm wavelength, unlike the 1525 nm wavelength, no signal above background was seen using NaGdF4:Nd/NaGdF4 NPs when the tissue slices were thicker than 1 cm. It is worth mentioning that, in order to accurately compare the penetration depth of 1525 nm and 1060 nm signals, the emitted power of the rare earth pellet and the output power of the NIR source were first matched to have identical spectral intensity before the tissue phantoms were applied.
In addition to good penetration depth, low detection threshold concentration with high resolution is also an essential quality for an optical biomedical imaging agent. As a proof-of-concept experiment, we embedded different concentrations (1, 5, 20, 50, 100 nM) of the amphiphilic 1,2-distearoyl-sn-glycero-3-phosphoethanolamine-N-[carboxy-(polyethylene glycol)-2000] (DSPE-PEG2000-COOH) modified C/S1/S2/S3 NPs and NaGdF4:Nd/NaGdF4 NPs into pork muscle tissue at varied depths (5, 8, 15 mm) to investigate the feasibility of bioimaging by a InGaAs camera. It can be seen that the SWIR signals of C/S1/S2/S3 NPs were detectable under 5 mm of pork tissue even when the concentration was as low as 1 nM. For tissue thickness 8 mm and 15 mm, the detection threshold increased to 20 nM and 50 nM, respectively. Nevertheless, the detection thresholds of NaGdF4:Nd/NaGdF4 NPs were 20 nM and 50 nM for 5 mm and 8 mm, respectively, much higher than those of the C/S1/S2/S3 NPs. For 15 mm thickness, little signal could be detected when NaGdF4:Nd/NaGdF4 NPs were embedded.
To evaluate the capability of these modified C/S1/S2/S3 NPs for in vivo imaging, bioimaging experiments were carried out on nude mice and SD rats. At first, we designed the bioimaging experiment with two groups of 6 week old nude mice. For group 1, 0.2 mL of different concentrations (1, 5, 20, 50, 100 nM) of water-soluble C/S1/S2/S3 NPs were introduced to the stomach by gastric syringe. For group 2, the same dose of water-soluble NaGdF4:Nd/NaGdF4 NPs was used instead. After 20 min digestion, SWIR images were taken of both groups upon 800 nm irradiation (0.2 W cm−2) with appropriate filters. Bright 1525 nm SWIR signals were detected when C/S1/S2/S3 NPs were used (Figure 1.11c). Moreover, the 1525 nm SWIR signals were still detectable when the concentration of NPs was as low as 5 nM. Considering the depth of the stomach in nude mice (3–5 mm), these results were consistent with the detection thresholds obtained from pork tissue. By comparison, the SWIR signals from NaGdF4:Nd/NaGdF4 NPs were weaker and could be detected only at a high concentration (>50 nM). Thus, we demonstrated that the low detection threshold property of C/S1/S2/S3 NPs can be extended to in vivo imaging in a small-animal model. To further highlight the use of the C/S1/S2/S3 NPs for deep penetration and high-resolution bioimaging, we chose SD rats as an animal model. A similar gavage procedure was performed on the SD rats using C/S1/S2/S3 NPs. Figure 1.11e shows that these C/S1/S2/S3 NPs also exhibit bright fluorescence in the stomach and intestinal tract of the SD rats, with the detection threshold determined to be 100 nM. The outline of the stomach can be well imaged after dissection, suggesting our previous images were taken in the right position. In addition, the depth of the SD rat stomach was determined to be 0.8–1.2 cm, further demonstrating the good bioimaging performance of the C/S1/S2/S3 NPs with deep penetration and low detection threshold.
1.3.1 An Explanation: Absorption–Scattering Theory
For optical in vivo imaging, the tissue penetration depth generally depends on the absorption and scattering of the excitation and emission light. Bashkatov et al.117 have proposed a theoretical model to calculate the penetration depth:
where µa is the optical absorption extinction coefficient which depends on the wavelength (Figure 1.12a), µs′ (∼λ−w) is the reduced scattering coefficient, and δ is the resulting penetration depth. The exponent (w) depends on the size and concentration of scatterers in the tissue and ranges from 0.22 to 1.68 for different tissues (Figure 1.12b). From this theoretical model and our results, we can conclude that water absorption is stronger for 1525 nm than for 1060 nm radiation, but 1060 nm radiation suffers from more tissue scatter than 1525 nm radiation. Obviously, in our case, the reduced scattering coefficient µs′ was more important in determining the resulting penetration depth δ than the absorption extinction coefficient µa. We therefore suggest that the imaging performance of SWIR is significantly influenced by wavelength, which directly affects both µa and µs′. Moreover, which of the two plays the more important role cannot be ascertained on the basis of simulation analysis but depends on the application, in this case biomedical imaging.
1.3.2 NIR-IIa Window
In 2014, Dai et al.118 explored a new biologically transparent sub-window in the 1.3–1.4 µm wavelength range (the NIR-IIa region) and performed non-invasive brain imaging in this window using SWNTs which emit in the whole 1000–1700 nm NIR II window. They resolved cerebral vasculature with a high spatial resolution of <10 µm at a depth of >2 mm in an epifluorescence imaging mode. Compared to previous NIR II work, they found that the 1.3–1.4 µm NIR-IIa window for in vivo imaging can further reduce tissue scattering by rejecting photons with a wavelength <1.3 µm. The truly non-invasive nature and dynamic capability of NIR-IIa imaging could allow cerebrovascular imaging with high spatial and temporal resolution to follow biological processes in the brain at the molecular scale.
The emission of lanthanide ions is not only abundant in the NIR-IIa, but also more spectrally pure than the emission of SWNTs. In 2015, García et al. demonstrated how the use of the particular emission band at 1340 nm of Nd3+ ions in SrF2 NPs can be used for deep-tissue, autofluorescence-free, high-resolution in vivo imaging.119 In their experiment, they found that food autofluorescence extends up to about 1100 nm upon 808 nm excitation, and this kind of infrared autofluorescence was found to exist in all the different diets available in their animal center. Nevertheless it is important to note here that both the fluorescence intensity and spatial distribution within the pellets were found to be strongly dependent on the particular pellet analyzed. Concerning the possible origin of this infrared luminescence, previous studies dealing with the visible autofluorescence of animal food have correlated it with the presence of plant components (in particular with the presence of alfalfa in the food pellets). García et al. also claim that the infrared food fluorescence they reported (extending up to 1100 nm) was generated by the presence of plant components, in particular by chlorophyll.
Up to this point, it is clear that high-contrast (autofluorescence-free) NIR in vivo imaging requires the use of luminescent materials that show intense fluorescence in the NIR-IIa region. The emission spectrum of Nd3+ (after 808 nm excitation) displays three emission bands centered at around 900, 1060, and 1340 nm corresponding respectively to the 4F3/2 → 4I9/2, 4F3/2 → 4I11/2 and 4F3/2 → 4I13/2 intra-4f electronic transitions of this ion. As mentioned above, the use of 900 nm and 1060 nm bands directly for fluorescence contrast have been reported. Nevertheless, it is clear that the use of these two bands does not ensure the complete removal of food autofluorescence, and these two bands are outside the NIR-IIa region. In fact, the removal of food autofluorescence would only be achieved if fluorescence images were recorded based on the 1340 nm fluorescence band. The principal reason why this band of Nd-doped NPs has not been previously used for in vivo imaging is its lower intensity when compared to the 900 and 1060 nm emission bands. Indeed, for most of the Nd3+-doped crystals the fluorescence branching ratio of the 4F3/2 → 4I13/2 (1340 nm) transition is about three times lower than that of the 4F3/2 → 4I9/2 (900 nm) or 4F3/2 → 4I11/2 (1060 nm) transitions. In fact, generally speaking, only 15% of the radiative de-excitations generated from the 4F3/2 metastable state of Nd ions is produced through the 4F3/2 → 4I13/2 fluorescence channel. This implies that using the 900 or 1060 nm bands would provide brighter images than those obtained from the 1340 nm emission. Thus, in vivo fluorescence imaging based on the 1340 nm weak emission would require the use of a host matrix providing large radiative de-excitation probabilities for Nd3+ ions. In this respect the SrF2 host seems to be particularly interesting as, compared to other fluoride nanocrystals, it has been shown to provide the highest emission intensities for other trivalent rare earth ions. This superior performance is based on various causes. First of all, multiphonon relaxation is made inefficient in this host due to the low wavenumbers of the vibrational modes. Moreover, although the local site symmetry at the cationic sites is highly symmetric in the SrF2 lattice (Oh), due to the necessity of charge compensation (Nd3+ replaces Sr2+), non-centrosymmetric crystal field components are expected to be present around the Nd ions, thereby leading to partially allowed forced electric dipole transition and, hence, to bright fluorescence. Thus, in principle, Nd:SrF2 NPs are expected to provide a highly intense 1340 nm emission, compared to other previously studied Nd:NPs. The QY of the as-synthesized Nd:SrF2 NPs is estimated to be 0.9 ± 0.1. Indeed, this large QY can be now compared to that recently reported for Ag2S QDs (QY = 0.15), which have also been used for high-brightness infrared fluorescence in vivo imaging.
García et al. then performed different infrared fluorescence imaging experiments by using an 808 nm diode laser as the excitation source and a Peltier-cooled InGaAs infrared camera for detection. Two different experimental configurations were adopted. In the “free-running” mode (FRM) no filters were used, such that the obtained fluorescence images accounted for the spatial distribution of emitted light integrated in the 900–1500 nm range. Fluorescence images were also obtained in the “food fluorescence free” mode (FFFM), by attaching a 1300 nm long-pass filter to the camera objective. Thus, in this mode images were constructed by recording only the 1300–1500 nm fluorescence range. The complete removal of food fluorescence by using the FFFM configuration is demonstrated in Figure 1.13. Figure 1.13a shows an optical image of a food pellet and an Eppendorf tube partially filled with the aqueous solution of Nd:SrF2 NPs. As can be observed in Figure 1.13b, both the food pellet and Nd:SrF2 NPs appear in the FRM fluorescence image. This agrees well with the emission spectra included in an ex vivo imaging experiment, where it is evident that a spectral overlap between food fluorescence and the 4F3/2 → 4I9/2 or 4F3/2 → 4I11/2 emission bands of Nd3+ ions occurs. In contrast, Figure 1.13c shows how the contribution of food emission to the fluorescence image was completely removed when images were acquired in the FFFM mode. The suitability of the 1340 nm emission of Nd:SrF2 NPs for autofluorescence-free in vivo and ex vivo imaging is demonstrated in Figure 1.13e–i. Figure 1.13(e and f) shows the fluorescence images of a nude mouse (Figure 1.13d) after intravenous injection of 50 µL of a phosphate buffered saline (PBS) solution containing Nd:SrF2 NPs (at a concentration of 0.3 wt%), as obtained in the FRM and FFFM configurations, respectively. Both images were obtained 1 h after injection using an excitation intensity 0.5 W cm−2 at 808 nm. The FRM fluorescence image (Figure 1.13e) reveals a bright luminescence generated from the whole abdominal and lumbar zones. On the other hand, the FFFM image (Figure 1.13f) provides better resolution and greater contrast, showing fluorescence only from a more localized zone at the upper area of the abdomen. Due to the complete absence of autofluorescence background, Figure 1.13f suggests a strong accumulation of Nd:SrF2 NPs in the liver and/or spleen. This point has been further corroborated by performing ex vivo experiments. Figure 1.13(h and i) shows fluorescence ex vivo images of the same mouse after being killed, as obtained in FRM and FFFM configurations, respectively. Figure 1.13i confirms that infrared fluorescence was mainly generated from the liver and spleen, without a relevant contribution from other organs such as lungs or kidneys.
1.4 Upconversion and Downconversion Dual-Mode Luminescence in One Nanoparticle
As is mentioned in the two previous sections, lanthanide-based NIR nanomaterials, both UCNPs and DCNPs, have become a specific topic of interest in recent years. Up to now, numerous lanthanide-based NIR nanomaterials have been developed, which can emit strong UC or DC luminescence by tuning different lanthanide dopants. However, it remains difficult to realize efficient nanoprobes with combined UC and DC dual-mode functions under a single NIR excitation and fulfill the excitation/emission wavelength requirements of in vitro and in vivo applications at the same time. Furthermore, for the UC process, because of well-established efficient UC luminescence, considerable efforts have been devoted to the synthesis of lanthanide-doped NaYF4(NaGdF4) NPs, where Yb3+, acting as the sensitizer with a large absorption cross-section around 980 nm, is usually codoped along with the most common UC activator ions (Er3+, Tm3+, Ho3+, Pr3+, and Tb3+) to produce strong visible and UV emissions. However, excitation light around 980 nm suffers from strong water absorption, as already discussed in the section on Nd3+-sensitized UCNPs. Therefore, if one can engineer nanomaterial structures with both the above-mentioned Yb3+–Nd3+–RE3+ UC system and the Nd3+-doped DC system, a Nd3+-sensitized UC/DC dual-mode nanoprobe under a single excitation around 800 nm can be realized, with a low heat effect and highly efficient bioimaging function. In 2014, our group developed a strategy to fabricate uniform multilayer C/S1/S2/S3 β-NGdF4:Nd/NaYF4/NaGdF4:Nd,Yb,Er/NaYF4 NPs that were composed of the NaGdF4:Nd core (DC) and the NaGdF4:Nd,Yb,Er shell (S2) (UC) in order to achieve dual-mode luminescence under a single excitation around 800 nm.120 This kind of NP could serve as a multiplexed luminescent biolabel for both in vitro and in vivo bioimaging applications. As shown in Figure 1.14a and b, this NP consists of four parts, each part having a specific role and working together to fulfill the dual-mode luminescence.
Nd3+ has an intense absorption cross-section at around 800 nm and highly efficient NIR-to-NIR DC luminescence (QY ∼ 40%) could be obtained when an Nd3+ activator was used. Therefore, in our C/S1/S2/S3 NPs the NaGdF4:Nd was constructed as a core for emitting the NIR DC luminescence under 800 nm excitation. This NIR-to-NIR DC luminescence is ideal for in vivo applications with deep light penetration, low light scattering, and heat effect because both the exciting and the emission wavelength are located in the NIR biological window. For the UC luminescence, unlike the typical single-sensitizer (Yb3+) upconversion system, efficient 800 nm-excited NIR-to-visible UC emission of Er3+ is realized (in S2) by taking advantage of Nd3+ and Yb3+ ions as double sensitizers, which can be used for the efficient in vitro bioimaging with a low autofluorescence and reduced photo-damage effects. Furthermore, we also found that there is a competitive relationship between DC and UC due to the energy transfer, governed by the law of conservation of energy. Hence, in order to avoid energy transfer between the DC core and the UC S2, a NaYF4 host layer (S1) was inserted. To ensure the photostability of the UC (S2), a NaYF4 host layer (S3) was also fabricated as the outer layer of the particle to minimize sublattice defects and external deactivators, as it is well known that these energy traps can be a major deleterious factor in regard to luminescence of colloidal NPs.
Dual-mode UC/DC luminescence could be obtained by the cooperation these layers. Upon NIR excitation around 800 nm, the C/S1/S2/S3 NPs exhibit characteristic UC and DC emission peaks of Er3+ (525 nm (2H11/2 → 4I15/2), 540 nm (4S3/2 → 4I15/2), and 650 nm (4F9/2 → 4I15/2)) and Nd3+ (862, 892 nm (4F3/2 → 4I9/2)), respectively. Comparing the luminescence intensity evolution during our synthesis, the strongest UC emission was observed in NaGdF4:Nd/NaYF4/NaGdF4:Nd,Yb,Er/NaYF4 C/S1/S2/S3 NPs. This was about 20 times stronger than that of NPs without the passivation NaYF4 layer (S3). In contrast, NaGdF4:Nd/NaYF4 has optimal DC emission, which is about 1.2 times stronger than that of C/S1/S2/S3 NPs. This result is reasonable because the 800 nm light source could be partially blocked and absorbed by the UC layer before reaching the DC core. It should be noted that NaGdF4:Nd NPs can also exhibit very weak UC emission around 525 and 585 nm under 800 nm excitation, which can be attributed to the 2K13/2, 4G7/2 → 4I9/2, and 4G5/2, 2G7/2 → 4I9/2 transitions of Nd3+. This weak UC emission of Nd3+ can be neglected after coating with the NaGdF4:Nd,Yb,Er/NaYF4 UC layers.
To determine whether these C/S1/S2/S3 NPs can be used for cellular imaging by using 800 nm-excited green UC emission, we have performed in vitro cellular bioimaging using human lymphocytes (as a cell suspension). After incubation with 0.2 mg mL−1 NaGdF4:Nd/NaYF4/NaGdF4:Nd,Yb,Er/NaYF4 C/S1/S2/S3 NPs in PBS (pH 7.4) for 3 h at 37 °C, the unbound NPs were washed away, and the living cells were imaged using 800 nm excitation. Cellular uptake of the NPs can be clearly observed from the merged bright-field and UC signal of cells. Local spectral analysis of overall cell staining confirms that our NPs are the origin of the cellular luminescence signal.
As a proof-of-concept experiment, we embedded the modified NaGdF4:Nd/NaYF4/NaGdF4:Nd,Yb,Er/NaYF4 C/S1/S2/S3 NPs into pork muscle tissue at varying depths (0–15 mm) to investigate the feasibility of bioimaging by a modified in vivo imaging system. As shown in Figure 1.14c, the NPs can be detected even at a depth of 15 mm under an excitation power density of approximately 1 W cm−2. In identical experimental settings, however, when traditional NIR (980 nm)-to-NIR (808 nm) UC NaGdF4:Yb,Tm/NaYF4 was used as a biomarker, the signals are much weaker than that of the C/S1/S2/S3 NPs at the same tissue depth, especially in deep muscle tissue (>10 mm). To demonstrate the capability of the NIR (800 nm)-to-NIR (860–895 nm) DC for in vivo imaging, 0.2 mL of 5 mg mL−1 water-soluble NPs was introduced to the stomach of nude mice by gastric syringe. A clear high-contrast luminescence image was observed, with almost no autofluorescence observed elsewhere (Figure 1.14d). It is noteworthy that the DC signal from C/S1/S2/S3 NPs can be detected even from the back of the mouse, suggesting that the 800 nm-excited low-heat-effect UC and DC dual-mode nanoprobe can not only be used for NIR (800 nm)-to-visible (540 nm) in vitro bioimaging but also shows great penetration capability in the NIR I window.
Within a year, the same idea was also shared by other groups, such as Qin et al. who reported sub-10 nm BaLuF5:Yb3+,Er3+@BaLuF5:Yb3+ NPs which can upconvert and downconvert 980 nm light simultaneously.121 Chaudhuri et al. designed and synthesized NaGdF4:Nd3+, Yb3+, Tm3+ nanophosphors with combined dual-mode DC and upconversion UC PL upon 800 nm excitation. A broad range of PL peaks covering NIR I/NIR II (860–900, 1000, and 1060 nm), and visible emission including blue (475 nm), green (520 and 542 nm), and yellow (587 nm) light were exhibited by NaGdF4:Nd3+, Yb3+, Tm3+ NPs after excitation at 800 nm.122 These results prove that such dual-mode luminescence NPs can open the door to engineering the excitation and emission wavelengths of UC/DC NPs and provide a new tool for a wide variety of applications in the fields of bioanalysis and bioimaging.
In the last decade, NIR biomedical imaging based on lanthanide nanomaterials has played an important role in biotechnology due to its intrinsic advantages over the traditional imaging probes. They benefit from high photostability, non-blinking properties, and low toxicity, but they also suffer from shortcomings. The absorption coefficient of UCNPs is quite low, which results in a low QY (<1%). As far as we know, the low absorption coefficient is limited by the low 980 nm absorption cross-section of Yb3+, which is even lower in the case of Er3+ and Tm3+. As mentioned above, many efforts have been made to improve the QY of UCNPs, such as changing the excitation wavelength. However, further improvements are still needed. There is only a handful of lanthanide DCNPs, which are not even well studied yet. Among the reported lanthanide-based DCNPs, the QY of Nd3+ is much higher than any other lanthanide ions such as Er3+, Tm3+, and Pr3+. More advanced nanostructures must be designed to reduce the crystal defects or modify the crystal field of the host. What is more, in order to obtain highly efficient lanthanide nanoprobes, increasing the doping concentration without lowering the QY is also something to think about.
The work was supported by the NSFC (Grant No. 21322508, 21210004) and the China National Key Basic Research Program (973 Project) (No. 2013CB934100, 2012CB224805).