Super-resolution upconversion microscopy of praseodymium-doped yttrium aluminum garnet nanoparticles

We demonstrate background-free subdiffraction optical microscopy with upconverting YAG nanoparticles doped with trivalent praseodymium ions ($Pr:\text{YAG}$). The presented microscopy with $Pr:\text{YAG}$ nanoparticles takes advantage of three facts. First, while excited with visible laser light, $Pr:\text{YAG}$ nanoparticles can emit upconverted ultraviolet (UV) radiation, thus allowing for background-free microscopy. Second, the technique based on exploiting donut-shaped laser beam was introduced to obtain subdiffraction-limited optical resolution. All optical resolution of 50 nm limited by the size of the particles was achieved. Third, $Pr:\text{YAG}$ nanoparticles are absolutely photostable. This technique resembles stimulated emission depletion microscopy (STED) though it is significantly different from STED since it involves stimulated absorption rather than stimulated emission.

Figure 1

Energy level diagram of $P{r}^{3+}$ electronic states in $\text{YAG}$ crystal. Two ways of stepwise excitation of upconverted UV fluorescence are shown. The excitation scheme used in the present work is indicated by bold arrows. It is also indicated that the second excitation step taken with a 609 nm laser hits the very edge of absorption into the $4f5d\left(1\right)$ state and that a 532 nm laser cannot serve as the first excitation step (indicated by dashed arrows).

Figure 2

Confocal/super-resolution microscope used in the present study. The confocal studies of $Pr:\text{YAG}$ nanoparticles were performed only with gaussian orange and green readout beams. Time-domain measurements were performed with all three beams being present, but with vortex plate taken out.

Figure 3

(a) Confocal image of $Pr:\text{YAG}$ nanoparticles. (b) Super-resolution image of the same area. The parameters of the pulse sequence are as follows: 1) the total sequence duration was $\approx 6.8\mu s$ corresponding to the repetition rate of 147 kHz, 2) the orange pulse duration was $\approx 1.3\mu s$ with the unchopped orange power being $\approx 5$ mW resulting in $\approx 1$ mW of average orange pump power, 3) the delay between the green readout and the beginning of the consecutive orange pulse was $\approx 240$ ns with the average readout power being $46\mu W$, 4) the power in the donut-shaped 532 nm depleting beam was $\approx 25$ mW, 5) the readout gate was open for $\approx 80$ ns starting $\approx 10$ ns prior to the readout pulse arrival, and 6) dwell time per pixel was 3 ms. (c), (d) Confocal and super-resolution scans of a nanoparticle cluster. The shape of the nanoparticle aggregate can be retrieved. Images (a) and (c) were obtained under the same conditions as (b) and (d) only with the donut beam blocked.

Figure 4

(a) Sequence of laser pulses used to evaluate the efficiency of intermediate state depletion by a 532 nm laser. (b) Time traces of fluorescence signals corresponding to the following powers of depleting beam: no depleting beam, 1 mW, and 29 mW at the input aperture of the objective lens. The right inset shows the depletion rate as a function of laser power while the left one demonstrates the decrease of signal produced by the readout pulse as the depleting beam becomes stronger.

Figure 5

(a) Confocal image of a single nanoparticle. (b) High resolution image of the same nanoparticle. The power of the donut beam used was 29 mW. (c) The dependence of the Gaussian half width of high resolution fluorescent spot as a function of donut beam power. The lower curve represents $1/\sqrt{P}$ fit of the low donut power data points. The resolution dependence clearly deviates from $1/\sqrt{P}$ for high $P$. (d) AFM image of the same particle revealing its $\approx 33$ nm height and $40\text{--}50$ nm lateral size.