1. Introduction

The rapid advances of the multiphoton microscopy (MPM) technique have been paving a new way to biological exploration and clinical diagnostics since it was invented in 1990 [1]. MPM enables anti-Stokes fluorescence with the fluorophore simultaneously absorbing two or more photons of longer wavelength. These nonlinear interactions are confined to the focal spot volume due to the power-dependent multiphoton emission $I_{fluo}\propto 〈I{\left(t\right)}^n〉$ [2]. The inherent optical sectioning ability has promoted the wide applications of MPM in biological exploration, such as cellular function, protein motion and tissue dynamics [3–6]. In contrast with confocal microscopy, MPM avoids photochemical damage of biological sample and increases the imaging depth in turbid specimens by utilizing excitation light of longer wavelengths [7–9]. Because of the indispensable high photon fluxes, the commonly used laser source in MPM is near infrared (NIR) femtosecond pulsed laser with an extremely high peak power. However, the high cost and high complexity of MPM system hindered it to enter into more laboratories.

Theoretically, the full width at half maximum (FWHM) of the intensity point spread function (IPSF) for the N-photon microscopic imaging could be described by the formula $\Delta x\Delta y\approx {1.22\lambda /2NA(N)}^{1/2}$ $(N\ge 2)$ [2], where $\lambda$ donates the wavelength of the excitation laser. Compared to confocal microscopy, the high-order-assisted high resolution advantage of MPM is suppressed by the long NIR excitation. In addition, the most studied multiphoton process including two-photon fluorescence (TPF) and second harmonic generation (SHG) are low-order nonlinear processes (N = 2), which cannot produce significantly high resolution [10]. According to the above formula, higher resolution requires for higher-order multiphoton emission and shorter wavelength of excitation light. However, conventional multiphoton imaging suffers from other problems. On one hand, shifting the excitation wavelength from the NIR (780 nm-2500 nm) to the visible range (380 nm-780 nm) [11], the corresponding high order multiphoton emission would be likewise shifted from the visible to the UV band (100-380 nm), which could hardly be detected by the commonly used detectors as well as have ultrahigh attenuation in the biosamples. On the other hand, as for traditional fluorophores the high-order multiphoton nonlinearity has extremely low efficiency and small absorption cross section, and the employed high power femtosecond laser is likely to severely damage the biosample. Therefore, it is really meaningful but challenging to achieve low peak power short-wave laser excited high-order multiphoton visible fluorescence microscopy.

Lanthanide-doped upconversion nanoparticles (UCNPs) are a novel class of luminescent nanomaterials for biological applications [12]. The sensitizers of UCNPs, typically Nd³⁺ and Yb³⁺, can absorb multiple excitation photons via intermediate energy states, while the activators convert the excitation energy into multiphoton luminescence [13, 14]. According to previous studies, the multiphoton emission efficiency of UCNPs is several orders of magnitude higher than that of many other multiphoton probes. In addition, the application of UCNPs also benefits from other advantages, including non-photobleaching, non-photoblinking, autofluorescence-free, high resolution, deep tissue and low cytotoxicity bioimaging with upconverting emission [15–20]. MPM using Yb³⁺-sensitized UCNPs (Yb³⁺-UCNPs) has been investigated [21, 22]. In addition, Liu et al. demonstrated that the improvement in the spatial resolution by using multi-photon upconversion emissions was theoretically and experimentally investigated [23]. However, the employed 980-nm-laser excited NIR-to-visible multiphoton process would induce an overheating effect to biological samples as well as that the long excitation wavelength would sacrifice the resolution [24]. Recently, compared with the Yb³⁺-UCNPs, 800-nm band light excited Nd³⁺-UCNPs was proposed with minimum heating effect in bio-system [25–28], which means that Nd³⁺-UCNPs are more suitable for biological imaging than Yb³⁺-UCNPs. In our previous studies, compared with Yb³⁺-UCNPs under 975-nm excitation, Nd³⁺-UCNPs under 800-nm-excitation could effectively improve MPM resolution [13]. According to the ladder-like energy levels of Nd³⁺, it is likely that light of shorter wavelength can effectively excite the ground state of Nd³⁺. However, there is no report on using visible light laser to excite Nd³⁺-UCNPs for high-order nonlinear emission.

In this work, we for the first time propose a visible-to-visible four-photon high resolution microscopic imaging using a visible CW laser to excite the prepared Nd³⁺-UCNPs. A common low-cost 730-nm laser diode was used as the laser source to build multiphoton scanning microscope system. The ultrahigh resolution of Nd³⁺-UCNPs using this novel excitation wavelength was systematically investigated in the two-photon, three-photon and four-photon microscopy. We also investigated the saturation power of three-photon and four-photon processes.

2. Results and discussions

According to previous reports, the luminescence mechanisms of Tm³⁺ doped Nd³⁺-UCNPs was shown in Fig. 1(a) [29]. It has provided potential to construct a visible-to-visible four-photon microscope system using 730-nm laser. Under the 730-nm laser excitation, Nd³⁺-UCNPs can generate multiphoton visible emissions. The energy transfer process of Nd³⁺-UCNPs under 730-nm laser excitation has been elaborated as follows. To begin with, Nd³⁺ ions absorb 730-nm photon energy in ⁴F_7/2, and then populate ⁴F_3/2 by sequential multi-phonon non-radiative relaxation. The bridging ion, Yb³⁺ ion, transfers the excited energy from sensitizer Nd³⁺ to the activator Tm³⁺. The energy levels of Tm³⁺, ³H₅, ³F₂, ¹G₄ and ¹D₂, are mainly populated by the energy transfer upconversion (ETU) process. The optical transitions corresponding to ¹G₄→³F₄, ¹G₄→³H₆, ¹D₂→³F₄ can generate luminescence centred at 650 nm, 474 nm and 455 nm, respectively, in two-photon, three-photon and four-photon processes [26, 29]. According to previous works, an optimized design of Nd³⁺-UCNPs, the efficient core-shell Nd³⁺-UCNPs (NaYF₄:30%Yb³⁺,1%Nd³⁺,1%Tm³⁺@NaYF₄:20%Nd³⁺) were successfully synthesized in our experiment [26, 29]. Specially, the doping ions and concentration was optimized for the four-photon imaging experiment. As shown in the Fig. 1(b), the morphology of the nanoparticles was characterized using the transmission electronic microscopy (TEM), indicating an average size of about 30 nm. The absorption spectrum of the as-synthesizedNd³⁺-UCNPs was measured and showed in Fig. 1(c). As indicated in Fig. 1 (c), 795-nm laser is not the short wavelength limit of the Nd³⁺-UCNPs excitation spectrum. Another strong absorption band around 730 nm of Nd³⁺ provides potential of using 730-nm visible laser to function as the efficient excitation source. Furthermore, as shown in Fig. 1(d) the four-photon emission peak centred at 455 nm of Nd³⁺-UCNP is relatively strong. In addition, the decay time of fluorescence from the ¹D₂→³F₄ and ¹G₄→³H₆ transitions were also measured by a time-correlated single photon counter (TCSPC) [30], as shown in Figs. 1(e) and 1(f). The lifetime of ¹G₄→³H₆ transitions was 110 μs, corresponding to three-photon 474 nm fluorescence. The lifetime of ¹D₂→³F₄ transitions was 78 μs, corresponding to four-photon 455 nm fluorescence. Furthermore, according to previous reports [31], there are many methods to shorten Nd³⁺-UNCPs’ fluorescence lifetime to increase imaging speed, including decreasing the size of UCNPs or improving the concentration of doped Tm³⁺. What’s more, streaking artifacts also could be outcome by adopting confocal pinhole [32] or combining with special algorithm [33]. Therefore, Tm³⁺ doped Nd³⁺-UCNPs can be used to improve lateral resolution in multiphoton imaging based on the theoretical analysis and spectral and fluorescence lifetime results.

 

Fig. 1 (a) The proposed multiphoton luminescence mechanism of Nd³⁺-UCNPs under 730-nm excitation; (b) the TEM image of the prepared Nd³⁺-UCNPs; (c) the absorption spectrum of Nd³⁺-UCNPs; (d) the emission spectrum of the prepared Nd³⁺-UCNPs with 650 nm, 474 nm and 455 nm peak; (e) lifetime of three-photon fluorescence from the ¹G₄→³H₆ transition; (f) lifetime of four-photon fluorescence from the ¹D₂→³F₄ transition.

Download Full Size | PDF

The home-built optical system for CW laser excited visible-to-visible four-photon microscopy is shown in Fig. 2. The CW excitation light source was the commonly low-cost 730-nm diode laser (Thorlabs HL7302MG). The beam passed through 715-nm longpass (Semrock FF01-715/LP-25) filter and a galvanometer system. After being reflected by a 670 dichroic mirror (Semrock FF670-SDi01-25x36) and directed into a high numerical aperture (NA = 1.4) oil immersion objective (Olympus 60x/oil), the excitation laser beam was focused on the sample. The emissions were backforward captured by the same objective. A 650-nm shortpass (Semrock FF01-665/SP-25) filter was used to further block the excitation light between the objective and reflection mirror (M3). After being reflected by M3, the fluorescence light was directed into a photomultiplier tube (PMT) operating in the photon counting regime and fluorescence images were obtained by the scanning of the studied sample with a galvanometer system on the computer (Olympus IX81). The emissions can be selected by adequate interference filters chosen according to the emission spectrum, including three-photon (Olympus BA460nm-500nm), four-photon (Semrock FF01-435/40-25). Without silver mirror (M3), the fluorescence light was collected by a spectrometer (QE65Pro, Ocean Optics) for spectral analysis.

 

Fig. 2 The home-built optical system for a CW laser excited visible-to-visible four-photon microscopy. F1: 715-nm longpass Filter, P1: half wave plate. P2: polarizing film. GM: galvanometer system, DM: 670 -nm dichroic mirror. F2: 665-nm shortpass emission filters. F3: adaptable bandpass filter. FL: focus lens. OL: objective lens. S: sample. PMT: photomultiplier tubes. M1, M2, M3: sliver reflection mirrors

Download Full Size | PDF

Next, the fluorescence imaging of Nd³⁺-UCNPs were obtained and analyzed in order to confirm visible-to-visible four-photon emission. Firstly, the synthesized Nd³⁺-UCNPs were sparsely spin-coated on a coverslip and characterized by scanning electron microscope (SEM) before optical imaging. The Nd³⁺-UCNPs were well dispersed in these samples, and a single Nd³⁺-UCNP was marked on the coverslip, as shown in Fig. 3. This single particle was excited by 730-nm CW laser in the home-built optical system (Fig. 2). Three milliwatt power (730-nm laser focal power density: 3.6 × 10³ KW/cm²) was employed to obtain two-photon image (overly saturated power). 30 μW power (730-nm laser focal power density: 35 KW/cm²) was employed to obtain three-photon image and four-photon image, respectively. At the same time, it is noted that the three-photon imaging was obtained at the speed of 200 μs/pixel and the four-photon imaging was obtained at the speed of 100 μs/pixel based on above fluorescence lifetime. At this scan rate, scanning and detecting with a PMT hardly lead to streaking artifacts, as shown in Fig. 4. For the three-photon and four-photon processes, excitation power needs to be controlled at the unsaturated level. When excitation power is higher than the saturation power, the resolution of these multiphoton processes will be reduced. For instance, the four-photon process can be converted to three-photon or even two-photon processes under overly saturation excitation power. The resolution will be reduced correspondingly. Therefore, it is necessary to keep excitation power unstructured for ensuring four-photon process in the experiment, which also means increasing the saturation power would benefit the experiments. Under different power densities, two-photon, three-photon and four-photon images were achieved to evaluate these FWHM_IPSF. Previous research indicates that in the process of multiphoton the intensity point spread function becomes (IPSF)^N, where N modifies the exponent of the Gaussian IPSF. When calculating the new FWHM_IPSF for a non-linear process of order N, the following formula for the instrument’s resolution is obtained:

 

Fig. 3 The SEM image of the highly dispersed Nd³⁺-UCNP sample: the imaged diameter of one single nanocrystal is 32.39 nm

Download Full Size | PDF

 

Fig. 4 (a) Two-photon (overly saturated three-photon excitation) fluorescence imaging of a single Nd³⁺-UCNP; (b) Three-photon fluorescence imaging of a single Nd³⁺-UCNP; (c) Four-photon fluorescence imaging of a single Nd³⁺-UCNP; (d) Gaussian fitting of two-photon fluorescence spot, FWHM_IPSF = 250 nm; (e) Gaussian fitting of three-photon fluorescence spot, FWHM_IPSF = 185 nm; (f) Gaussian fitting of four-photon fluorescence spot, FWHM_IPSF = 161 nm.

Download Full Size | PDF

The optical resolution achieved by imaging with upconverting emissions was quantified by measuring FWHM_IPSF. According to Eq. (1), the FWHM_IPSF are calculated and equal to 224 nm, 183 nm and 159 nm for the cases of N = 2, 3 and 4, respectively. The experimental results are also deduced by fitting to one-dimensional Gaussian probability to single-point intensity profiles. The calculated results of corresponding fluorescent spot were shown in the Fig. 4. The FWHM_IPSF of images were calculated to be 250 nm, 185 nm, and 161 nm for the two-photon, three-photon and four-photon process, respectively. Apparently, these results are in good agreement with the theoretical prediction. According to the later data processing and image processing, compared with the use of two-photon emission. the imaging resolution could be apparently increased by a factor of 1.56 through using four-photon emissions, Therefore, the resolution could be improved by using four-photon emission and it will be still valuable for discerning more detailed structures.

To further confirm the four-photon upconversion mechanism under 730-nm irradiation of Nd³⁺-UCNPs, the power dependence of emission intensity was measured for the high-order processes. The correlation between luminescence intensity and excitation power can be described by the formula $I\propto P^n$where I represents the emission intensity, P donates the pumping laser power, and n is the number of photons involved in the excitation process [34]. Parameter n could be derived by measuring the emission images in the microscope system and its corresponding pump laser power. Figure 5 shows the experimental dependence of the two different luminescent emission bands versus the incident excitation power. Multi-photon fluorescence intensities were chose by appropriate filter according to multi-photon emission band and intensities were collected photomultiplier tubes (PMT). These fluorescence intensities were measured from single Nd³⁺-UNCP. Indicated by the curves, the saturation power of these UCNPs was relatively high for both three-photon and four-photon emission. In the low power range, a 3.47 slope of 455-nm luminescence was achieved, indicating a four photon emission process. A 2.63 slope of three-photon 470-nm luminescence was also obtained. In previous study, four-photon imaging of 980-nm laser excitation Yb³⁺-UCNPs was obtained by utilizing 5 μW power. However, four-photon imaging of 730-nm laser excitation Nd³⁺-UCNPs was obtained by utilizing 30 μW average power. What’s more, compared with saturation power, the saturation power Nd³⁺-UCNPs four-photon imaging was estimated to be about 40 μW, which is 8 times higher than that of 980-nm laser excitation Yb³⁺-UCNPs four-photon emission [21].

 

Fig. 5 The relationship between the multiphoton fluorescence intensity and excitation saturation power in NaYF₄: Nd³⁺, Yb³⁺, Tm³⁺.

Download Full Size | PDF

According to above theoretical analysis and experimental demonstration, the lateral resolution could be remarkably improved by 730-nm CW excitation. A sample with fine structure was imaged to demonstrate the advantages of visible-to-visible four-photon ultrahigh resolution microscopic imaging with 730-nm diode laser excited nanocrystals, as shown in Fig. 6. Nd³⁺-UCNPs solution was dropped on the coverslip as shown in Fig. 6(a). Fluorescence image of Nd³⁺-UCNPs was shown and the relatively bright spots were resulted from the aggregated Nd³⁺-UCNPs’ as shown Fig. 6(b). Nd³⁺-UCNPs was excited to obtain four-photon imaging under unsaturated 730-nm power level. In addition, two-photon images were also obtained under 730-nm saturated excitation (650 nm emission) and 980-nm excitation (650 nm emission), respectively. The scanning images for the zoomed in area were shown in Fig. 6(c)-6(e), in which the aggregated Nd³⁺-UCNPs appeared with fine structure. Compared with two-photon of 980-nm excitation, Fig. 6(c)-6(f) indicated that the improved resolution can help discern fine structure better by two-photon and four-photon using 730-nm laser. It is noted that more detailed structures utilizing four-photon were shown in Fig. 6(c). With analysis and calculation, FWHM = 168 nm, FWHM = 250 nm and FWHM = 360 nm were shown in Figs. 6(f) and 6(g), corresponding to four-photon of 730-nm, two-photon of 730-nm, and two-photon of 980-nm. Apparently, compared with two-photon using 730-nm excitation and 980-nm excitation, the resolution was improved by four-photon of 730-nm excitation. Furthermore, the four-photon process of Nd³⁺-UCNPs is so saturated easily that the resolution is reduced. Of course, further increasing the saturation excitation power would better popularize the proposed visible-to-visible four-photon ultrahigh resolution microscopic imaging.

 

Fig. 6 (a) bright field image and (b) UC fluorescence image of Nd³⁺-UCNPs; (c)-(e) High-magnification fluorescence images of aggregated Nd³⁺-UCNPs as marked in (b). (c) four-photon image by 730-nm laser excitation; (d) two-photon image by 730-nm laser excitation; (e) two-photon image by 980-nm laser excitation; (f) line profile of signal intensity of marked line in (c)-(e); (d)-(h) The corresponding line-scanning profile from the image shown in (f) showing FWHM = 168 nm (730-nm/four-photon); FWHM = 250 nm (730-nm/two-photon); FWHM = 360 nm (980-nm/two-photon).

Download Full Size | PDF

3. Conclusion

In conclusion, we have proposed and demonstrated a visible-to-visible four-photon microscope system using 730-nm laser to excite Nd³⁺-UCNPs with several advantages of UCNPs in multiphoton microscopy. Employing 730-nm laser the multiphoton scanning imaging system was built and the two-photon, three-photon, and four-photon imaging of single nanocrystal were investigated. High lateral imaging resolution up to 161-nm was achieved via the four-photon upconversion process. The demonstrated large saturation excitation power for Nd³⁺-UCNPs would be more practical and facilitate the experimental four-photon imaging in the applications. The visible excitation wavelength and high order optical processes lead to higher resolution and provides new opportunities for their application in multiphoton microscopy. A multiple nanoparticles aggregated UCNPs sample with fine structure was imaged to demonstrate the advantages of this proposed visible-to-visible four-photon ultrahigh resolution microscopic imaging. Although there are some challenges in experiment, including controlling excitation power, fluorescence intensity of four-photon and optimized optical system, all favorable features still open a way to high resolution, cost-effective multiphoton microscope systems using low-cost visible laser sources, under the comprehensive consideration.