Characterizing the luminescent properties of upconversion nanoparticles in single and densely packed state

1. Introduction

Photon upconversion (UC) is a process in which the sequential absorption of two or more photons leads to the emission of light at a shorter wavelength than the excitation,^1,2 such as the conversion of infrared light to visible light. Upconversion nanoparticles (UCNPs) received much attention in recent years for their potential application in biomedical imaging, cancer diagnosis, and therapy. Higher penetration depth, low auto-fluorescence and minimized photo-damage made the UCNPs a promising alternative probe to molecular fluorophore for biomedical imaging of cells, tissue, and even live animals.^3–7

In contrast to traditional two-photon absorption/emission process, a long-lived intermediate state usually exists in UCNPs which greatly increases the quantum efficiency. Hence, a low-peak-power continuous laser could be used to excite the UCNPs for scanning microscopy instead of expensive femtosecond pulse laser.^8,9 In 2009, Yu et al. reported laser scanning upconversion luminescence microscopy (LSUCLM) with a confocal pinhole for 3D section imaging.⁹ In 2015, Hodak et al. found that nonlinear dependence of luminescent intensity on excitation power could be realized using pulse excitation.⁸ By adjusting power and excitation rate, optical sectioning similar to traditional two-photon microscope was achieved.⁸ However, the nonlinearity was obtained on the relative dense sample. For application in cell biology, mostly the sample needs to be labeled with a low concentration of UCNPs yielding a single particle state. So characterizing the luminescent properties of UCNPs is demanded.

So far, luminescence properties of UCNPs were characterized mostly in ensemble measurement. To increase the photon efficiency of UCNPs by optimizing the synthesis strategies, i.e., changing the morphology, size, doping ions etc,^10,11 ensemble measurements are sufficient. But the underlying UC mechanism could be averaged out with a large number of particles.¹² To recover that, measurements need to be done at single particle level.¹³ In 2014, Gargas et al. characterized sub-10nm up-converting nano-crystals for single-molecule imaging.¹⁴ However, its UC process and the difference to bulk samples is still not clear.

In this paper, luminescent properties of single and densely packed UCNPs were studied with a custom-built laser scanning microscope. The dependency of the luminescent intensity of excitation laser power and time domain luminescent measurement demonstrates a single-photon adsorption process due to saturation of the intermediate state for single UCNPs. But a two-photon adsorption and a continuous luminescent increase even after turning off the excitation laser power was observed for densely packed UCNPs. The results suggest more intermediate states and excitation and decay pathways between neighboring particles exist in the densely packed sample.

2. Materials and Methods

Previously synthesized Er${}^{3+}$ and Yb${}^{3+}$ co-doped CaF₂ UCNPs were stored as dried powder. A 10mg powder was suspended in 1ml cyclohexane and shaken gently by hand. For the densely packed sample, 1 drop of 10μl sample was spread on freshly cleaned coverslip. For the sparsely dispersed single particle sample, the solution was further diluted 10,000x with cyclohexane and sonicated for 5min, then 1 drop of 10μl sample was spread on a freshly cleaned coverslip. Both samples were dried in air. 10μl water was dipped on the sample surface before mounting on the microscope.

A previously custom-built microscope was modified slightly to measure the UCNPs’ luminescence properties,¹⁵ as shown in Fig. 1 Briefly, a 980nm laser was used as excitation light source and a pockels-cell was used to turn on/off the light with sub $\mu$s time resolution. The laser beam was raster scanned by a pair of Galvo mirrors and guided into objective by a scan lens and a tube lens. A 60$\times$water-immersion objective (Nikon Plan Apo VC 60$\times$) was used for imaging and collecting emitted luminescence. The luminescent signal was separated from excitation laser by a dichroic mirror and further cleaned by a band-pass filter. A PMT detector and a spectrometer were used to detect the intensity and measure the spectrum, respectively, switched by a flipper mirror.

The microscope can work at three different modes: spectrum recording mode, scan imaging mode and time domain luminescent intensity measuring mode. When working at spectrum recording mode, the Galvo-mirrors remain at a fixed position and the flipper mirror was flipped up to guide the luminescent light into the spectrum. While working at scan imaging mode, the luminescent intensity was detected by the PMT with the Galvo-mirror raster-scanning the sample to record the 2D image. Once the $t$ image was recorded, points of interesting could be selected by hand, then the Galvo-mirror stopped the scan and kept at these positions one by one iteratively to record the time-domain luminescent intensity. At each position, the excitation laser was turned on by the electro-optic modulating pockels-cell for a short period of time, generally 500$\mu$s, and the luminescent intensity was recorded continuously until decay to zero.

3. Results and Discussion

3.1. Spectrum of UCNPs particles

Densely packed UCNPs sample was prepared by drying high concentration suspension on coverslip. When excited by 980nm laser with 1mW laser intensity, strong luminescence could be seen by naked eyes. The measured spectrum is shown in the inset of Fig. 1. The spectrum is ranged from 514nm to 550nm with maximum intensity at 541nm. There are another two weak emission peaks at 520nm and 532nm, respectively. This 550 band belong to ⁴S${}_{3∕2}$ to ⁴I${}_{15∕2}$ which is more suitable for kinetic studies.⁸ It is known that UCNPs has another emission band at 660nm, which corresponds to (⁴F${}_{9∕2}-{}^4$F${}_{15∕2})$, which is blocked by the dichroic mirror in our system.¹⁵ In the following experiment, one addition band-pass filter (500–560nm) was used to further clean the luminescent signal. With the same excitation laser intensity, the luminescent signal is too strong to be recorded by a PMT detector. So, in the following experiment, the detection of densely packed UCNPs sample was all done with the spectrometer.

3.2. Imaging UCNPs in single particle state

Single UCNPs sample was prepared by diluting the suspension 10,000$\times$ times and then drying on coverslip. The excitation laser was scanned across the sample by Galvo-mirrors and the luminescence intensity was detected by PMT. Figure 2 shows the luminescence image with different scanning speed. With a slow scanning speed of 50 lines/s, corresponding to 40$\mu$s dwell time for each pixel, most spheres can be clearly distinguished and have a round shape which verified the sparely dispersed single particle states. However, at fast scanning speed, the spheres appear to be elongated. The particles were elongated more and more with increased speed, i.e., shorter of dwell time.

The elongation at fast velocity was not due to the scanning mechanics of the microscope since the peak positions of each particle remain the same with different velocities. We hypothesize that they were due to the long lifetime of UCNPs. Since the microscope was working at nondescan mode, there was no pinhole used to reject the luminescent signal out of the excitation focus spot. In case of luminescence lifetime longer than the pixel dwell time, the excited UCNPs continue to emit luminescence even after the laser beam was scanned to the next pixel position. As a result, in the 2D image construction, the luminescence was also contributed to the next pixel position, hence the image was elongated. In order to avoid smearing, accurate imaging of single-particle UCNPs requires a slower scanning speed. In the following, a scanning speed with a dwell time of 40$\mu$s was used.

3.3. Dependence of luminescent intensity on excitation power

Next, we measured the luminescent intensity dependence on excitation power. For sparsely dispersed single particles, the sample was imaged at low scanning speed (40$\mu$s pixel dwell time) with different excitation laser power ranging from 100 to 1400$\mu$W. The intensity of each UCNP was calculated as the sum of a 13-pixel area around the particle after subtracting background. Figure 3(a) shows a plot of a typical UCNP which demonstrates a nearly-linear dependency. For the densely packed sample, the luminescent intensity at the peak of 545nm read from the spectrometer while varying the excitation laser power. On the contrary to single UCNPs, a nonlinear dependency is clearly seen for densely packed UCNPs as plotted in Fig. 3(b).

Generally, the dependence of luminescence intensity on excitation power can be described by a power-law equation³:

where $I_{\mathrm{\text{UCNPs}}}$ represents the luminescent intensity, $P$ represents the excitation power, and $m$ represents the absorption coefficient which is 1 for single-photon absorption and 2 for two-photon absorption process.

Insets of Fig. 3 show the log–log plot of the luminescent intensity dependent on the excitation power. By fitting the log–log plot with a linear function, the slop represents the absorption coefficient $m$. The absorption coefficients of 0.91 and 1.91 were obtained for single UCNPs and densely packed UCNPs, respectively. The absorption coefficients are not integer but fractional indicating complex process were involved during the UC. But it is clearly demonstrated that the single UCNPs undergo a more-or-less one-photon absorption process while densely packed UCNPs undergo a two-photon absorption process with the same 980nm laser excitation.

The different dependency between single and densely packed UCNPs can be explained by the saturation of the intermediate state. Different to general two-photon absorption with a femtosecond laser, one or more real long-lived intermediate states exist as energy reservoirs in the UC process. Upon absorption of the first photon, the particles undergo a transition from ground state to the intermediate state, stay there for up to hundreds of $\mu$s until absorption of the second photon or decay back to ground state. Upon absorption of the second photon by either excited state absorption (ESA) or energy transfer up-conversion (ETU) process, the particles transit from intermediate to the excited stated and are capable of emitting luminescence. For single UCNPs, although the particles need to absorb two photons to emit luminescence, the long-lived intermediate states prevent its decay to ground state before absorption of the second photon. In other words, the intermediate state always remains saturated upon the photon flux, so the dependence of the luminescent intensity on excitation power will only represent the transition process from intermediate state to excited state. Hence, a single-photon absorption behavior was observed as shown in Fig. 3(a). On the contrary, the absorption coefficient of 1.91 in Fig. 3(b) suggests that intermediate state is not saturated in densely packed UCNPs in the same range of excitation power. This could be due to more intermediate states or more decay pathways from the intermediate state to ground state for densely packed UCNCPs.

3.4. Time-domain luminescent intensity measurement

We further characterize the UC and luminescence process by time-domain measurement. Briefly, positions of each identified single particle were obtained from the image with low scanning speed. Then the excitation laser was turned off and scan mirrors were kept at each position successively so that the laser beam was focused on the particle for a certain period of time until the measurement is finished. At each position, a function generator generates a 500$\mu$s pulse to turn on the laser and trigger the acquisition of luminescent intensity simultaneously. For the densely packed sample, the same measurements were performed except that the focus position does not need to be changed. Figure 4 shows typical time-domain luminescent intensity curves for single UCNPs and densely packed sample.

The curves constitute two regions, the first 500$\mu$s region corresponds to UC with excitation laser on; the rest corresponds to luminescent decay with laser off. In the laser-off region, both samples exhibit single exponential decay although the decay time of a single particle is much shorter than the densely packed sample. The laser-on region is more complicated because it may involve a few photon absorptions and emission pathway at the same time. For single particles, the luminescent intensity first undergoes a fast increase ($<$100$\mu$s) then slowly reaches a plateau which suggests the saturation of excited states. For the densely packed sample, the luminescent intensity increased slowly at the beginning ($<$30$\mu$s) then continually increased in a nearly-linear behavior. Interestingly, even after the excitation laser was turned off, the luminescent intensity still increased for about 20$\mu$s for the densely packed sample, as shown in the inset of Fig. 4.

3.5. Upconversion model for single and densely packed UCNPs

In the above experiment, differents absorption coefficients and different time-domain luminescent behavior for single and densely packed UCNPs were observed. Gamelin and Gudel developed a simplified three-state model to simulate the process.¹⁶ We used the same model to describe the difference between single and densely packed UCNPs. Figure 5(a) shows the scheme of the electron state distribution of Er${}^{3+}$-Yb${}^{3+}$ co-doped UCNPs. Er${}^{3+}$ at the shell first absorb one 980nm photon and transit the molecule from ground state $S_0$ to intermediate state $S_1$. The energy can be transferred to the first excited state of Yb${}^{3+}$. Upon adsorption of the second photon, the particle was transited to emissive state $S_2$ from excited states $S_1$. $N_0$, $N_1$ and $N_2$ denote the populations of the ground ($S_0$), intermediate ($S_1$) and emissive ($S_2)$ states, then the absorption and emission process can be described as following equations⁸ :

where $G$ is the ground state excitation rate constant, $k_1$ and $k_2$ are the decay constant from the intermediate state and the emissive state to ground state, respectively, $E$ is the excited state excitation rate constant, by which absorption of the second photon can excite the particle from intermediate state $S_1$ state to emissive state $S_2$. The luminescent intensity is proportional to the population of the emissive state and the decay rate from emissive state to ground state, i.e., ($k_2\times N_2$).

Time-domain luminescent intensity curves in Figs. 4(a) and 4(b) were fitted to above function using the ode45 method. Table 1 lists the parameters of best fitting results to single and densely packed UCNPs. The results show that $G$ of densely packed UCNPs are 30 times larger than that of single particles, which suggests a strong cooperative absorption between neighboring particles in the densely packed sample. Because the ground states are mostly distributed in the shell, there are possible excitations from the ground of one particle to the intermediate states of neighboring particles. However, the ETU rate constant $E$ is about the same. This is because both the intermediate and the emissive states are mostly distributed in the core, hence there is less possibility of excitation from an intermediate state of one particle to the emissive state of another particle. The results also show that both the decay constants $k_1$ and $k_2$ of densely packed UCNPs are 2–3 times larger than that of single UCNPs, indicating multiple decay pathways exist for the former since it is possible from the intermediate and emissive state of one particle to the ground state of nearly other particles, as illustrated in Fig. 5(c).

The continual luminescent intensity increase after excitation laser is turned off is also presented in the fitted data for densely packed UCNPs, as shown in the inset of Fig. 4(b). This can be explained as that there are enough electrons at the long-life intermediate state, $E\times N_1$ is still larger than $k_2\times N_2$ which leads to the electrons at the emissive state still increasing even when the excitation laser is powered off during this short 20$\mu$s period. This phenomenon was not observed in single UCNPs indicating that the number of intermediate states is much lesser than compared to densely packed sample. So even the excitation rate constant $G_0$ is small for the intermediate state which is more likely to be saturated.

4. Conclusions

With a custom-built scanning microscope, luminescent properties of single and densely packed UCNPs were studied. Both luminescence-excitation dependency and the time-domain luminescent measurement suggests saturation of intermediate state for single UCNPs which leads to one-photon adsorption behavior. However, same measurements on densely packed UCNPs suggest that more intermediate states exist as well as more excitation and decay pathways between neighboring particles, which leads to two-photon adsorption behavior and increased excitation and decay rate constant. The very different luminescent properties of single UCNPs to densely packed ones will affect the experiment design and obtained results for its application in biomedical imaging.

Acknowledgments

The work was supported by the National Key Research and Development Program of China (YFC20170110100), the National Natural Science Foundation of China (NSFC) (Grant Nos. 61475185 and 11504409), and Natural Science Foundation of Tianjin City (TJNSF) (Grant No. 16JCYBJC43800). Xiaohu Chen and Zhengyu Gui contributed equality to this work.