1 Introduction

Quantum dots are semiconductor nanocrystals, 1–20 nm in size [1, 2], that possess a discrete energy level structure. The level structure is determined by the size, composition and shape of the quantum dot [3, 4]. Particular focus has been placed on II–VI semiconductors, such as CdSe [1]. By selecting for size, the absorption and fluorescence emission properties can be tailored to the application, which is attractive for implementations such as fluorescent labels [5, 6]. In addition, the quantum dot surface can be chemically functionalized to bind to specific molecules. Compared to fluorescent molecular markers, quantum dots possess high brightness, low bleaching rate and absorption over a wide wavelength range [6]. Similar to the fluorescence from other labels, quantum dot fluorescence can be driven by both single- and multiphoton processes [7–12].

Fluorescent labels are conventionally excited by short wavelength light to produce fluorescence emissions. Ultraviolet (UV) light is often used with quantum dots due to their absorption increasing toward shorter wavelengths [1]. For bioimaging applications, it is often important for the excitation light to penetrate far into the sample, but the penetration depth is limited by Rayleigh scattering. Since Rayleigh scattering scales as \(\lambda ^{-4}\), long-wavelength multiphoton excitation seems promising for imaging applications. Two-photon excitation of quantum dots is now routinely performed, often using Ti:sapphire lasers operating around 700–1,000 nm [11, 12]. In a recent study, Horton et al. [13] demonstrated in vivo three-photon microscopy with an Er-doped fiber system and red fluorescent protein labels. They found that an optimum wavelength window due to scattering and absorption in tissue lies close to the telecom band. Quantum dots are attractive for use in such applications due to their large three-photon absorption cross section [8, 9].

We demonstrate three-photon excitation and fluorescence in commercially available CdSe quantum dots using a custom built mode-locked Er-doped fiber laser (EDFL). Such lasers, and their associated optical components, are central to telecommunications and have therefore been the subject of extensive industrial research. These developments mean that they provide a stable, versatile and comparatively inexpensive method to generate ultrashort pulses for three-photon spectroscopy of quantum dots. We also excited single-photon fluorescence using a 372-nm diode laser to compare one- and three-photon fluorescence emission spectra, excitation intensity dependence and bleaching rate. Previous investigations have also demonstrated two- and three-photon excitation in quantum dots [10, 12]. However, to our knowledge, multiphoton excitation and absorption have not yet been investigated in quantum dots for excitation wavelengths in the telecom band.

The wavelength of expected fluorescence emissions from the sample is determined by the quantum dot energy level structure, which can be simplified to a ground state, fluorescent excited state and a continuum of absorption states [14, 15]. Upon excitation to this continuum of states, a quantum dot undergoes a non-radiative decay to the excited state, followed by fluorescent decay back to the ground state [14]. In addition, intense excitation pulses will interact nonlinearly with the sample to generate harmonics of the excitation wavelength. In quantum dots, the third-harmonic generation (3HG) process is expected to be the most significant of these [16]. The process of second harmonic generation (SHG) is negligible in comparison with 3HG due to the centrosymmetric crystal structure of CdSe [17, 18]. As the product of a coherent process, the harmonics are emitted in line with the excitation beam.

2 Method

Figure 1 shows the experimental setup used to probe the quantum dots. Colloidal Invitrogen™ Lumidot™ CdSe/ZnS 640 core-shell quantum dots, with a core diameter of 6.3 nm and a nominal fluorescence wavelength of 640 nm, were deposited on a sample slide. The toluene solvent was evaporated off, leaving a static quantum dot sample which was placed at the focus of the excitation beam. Collected fluorescence was coupled into a multimode fiber, where the spectrum was measured by a fiber-coupled spectrometer.

Fig. 1
figure 1

Experimental setup. Schematic for spectroscopy of quantum dots. a 372-nm light is focused onto the quantum dot sample and fluorescence is collected and coupled into a multimode fiber. Fluorescence is separated from residual excitation light using a Semrock FF511-Di01 dichroic mirror. b 1,575-nm pulses are focused onto the quantum dot sample and fluorescence is collected off-axis to avoid excessive coupling of third-harmonic (3HG) light. The majority of residual excitation and 3HG light continues to propagate on-axis past the slide and is not coupled into the multimode fiber (MMF). As some scattered light is still collected, KG3 colored-glass filters are used to block any remaining IR

Full size image

For the three-photon spectroscopy of the quantum dot sample, a mode-locked EDFL was used to produce 1,564-nm pulses at a repetition rate of 300 MHz. 116-fs pulses with a maximum average power of 186 mW were produced through Er-doped fiber amplification and soliton compression. The laser was focused to a spot with a \(14.1\pm 0.2\;\mathrm {\mu m}\) \(1/e^2\) radius, yielding an average intensity up to \(48.8\pm 0.5\;\mathrm {kW/cm}^{2}\). The soliton self-frequency shift (SSFS) [19, 20] of the amplified pulses shifted the center wavelength to 1,575 nm, increasing the spectral bandwidth to 85 nm. The pulse durations were measured by frequency-resolved optical gating (FROG) using a MesaPhotonics FROGScan with an Ocean Optics HR2000+ spectrometer, and a wavelength range of 700–881 nm. The spectrum was measured using a HP 70950A optical spectrum analyzer.

The resulting fluorescence emitted from the quantum dots was collected off-axis with a 35-mm focal length lens to spatially separate it from the large amount of light produced by 3HG. However, some of the 3HG and excitation light was scattered by the sample and collected. The fiber-coupled fluorescence was detected by an Ocean Optics USB650 spectrometer with a 350–1,000-nm wavelength range and 1-nm/pixel resolution. Three 2-mm-thick KG3 colored-glass filters were used to block any remaining IR, with a calculated attenuation of \(-80\) dB, in order to prevent damage to the detector.

The fluorescence detection spectrometer was used to capture all data for the fluorescence and 3HG spectra, and its dependence on the intensity of the excitation beam and exposure time. Measurements of the spectra were taken for different input intensities using a variable ND filter, and integration times from 5–30 s were used. The relative spectrum amplitudes were inferred from Gaussian fit profiles of the spectra.

To check that the quantum dots behaved as expected under UV excitation, we applied up to 3.5 mW of 372-nm light from a fiber-coupled Nichia diode laser to the quantum dot sample. Basic optical isolation was employed, using a quarter waveplate and linear polariser, to ensure that the operation of the diode laser was not compromised by optical feedback from the single-mode fiber. The UV light was focused to a spot with a 5.6 μm 1/e2 radius at the quantum dot sample, with a maximum intensity of 3.54 kW/cm2.

The fluorescence excited by the UV laser was collected by an \({\text {NA}}=0.60\) aspheric lens with a 4.02-mm focal length. This was separated from the residual excitation light by a Semrock FF511-Di01 dichroic mirror, with an edge wavelength of 511 nm, and coupled into MMF. An Ocean Optics USB4000 spectrometer, with a 316–1,021-nm wavelength range and 0.22-nm/pixel resolution, was used to measure the spectrum of the UV-induced fluorescence. Measurements were taken with varying input intensities and an integration time of 10 ms.

3 Spectra

Figure 2 shows the spectra from UV and IR excitation. From a Gaussian fit of the fluorescence spectra, the center wavelength from linear excitation is 632.6 ± 0.3 nm, where it is 631.4 ± 1.1 nm in the spectrum produced by three-photon excitation. The FWHM linewidths are determined to be 47.8 ± 0.3 and 40.7 ± 1.1 nm, respectively. The near-identical characteristics of the fluorescence spectra under UV and IR excitation indicate that the same excited energy band is reached in both cases. The increased linewidth under UV excitation could arise from power broadening and/or heating of the quantum dot sample under UV excitation.

Fig. 2
figure 2

Comparison of spectra from a λ = 372 nm excitation and b λ = 1,575 nm excitation, shown with the energy level structures and excitation pathways. In b, both the 640-nm fluorescence peak and a 540-nm 3HG peak are observed. c Spectrum of IR excitation beam on a linear scale, where the wavelength axis aligns at a ratio of 3:1 to that of b, showing the correlation of the 3HG peak with the fundamental. d FROG trace of IR pulse, where the wavelength axis refers to the second harmonic of the excitation wavelength

Full size image

In the three-photon excitation spectrum, the secondary peak at 534.8 ± 0.6 nm corresponds with the third harmonic of the IR excitation pulse. The peak is visible in the off-axis configuration due to scattering from the quantum dot sample. As shown in Figure 2, the third harmonic is generated predominantly from the red-shifted soliton at the long-wavelength end of the excitation pulse spectrum. This part of the excitation spectrum, according to FROG trace measurements, corresponds to the most intense part of the excitation pulse.

As a static sample of quantum dots was used for this experiment, the time-dependent spectra were also observed in each case, using an integration time of 500 ms over 10 min, to compare bleaching rates of the linear and three-photon excitation processes. Figure 3 shows the measured time-dependent intensity of fluorescence spectra under UV and IR excitation. We observed that although a higher excitation intensity was used for the three-photon process, the emission signal of the quantum dots was bleached to 50 % of the starting intensity after 566 s, compared to 103 s under linear excitation. However, the bleaching of the sample under IR excitation was not fit well by the expected double-exponential profile, suggesting that it may not follow a conventional bleaching mechanism. Further study is needed to properly assess the meaning of these preliminary observations.

Fig. 3
figure 3

Time-dependent intensity of fluorescence spectra under UV and IR excitation for 10 min. The vertical axis shows normalized intensity. Data from UV excitation are denoted by the blue data points, and data from IR excitation are denoted by the red data points

Full size image

4 Intensity dependence

To verify the photon orders of the two excitation processes and search for saturation of the excitation, we measured the dependence of the fluorescence on excitation intensity, as shown in Fig. 4. The fluorescence spectra for both the one- and three-photon measurements, taken at varying excitation intensities, were fit with Gaussian profiles. The amplitudes of the fit profiles were compared using power law fits, as shown in Fig. 4, in order to determine a \(F \propto I^{n}\) power law dependence between excitation intensity \(I\) and fluorescence counts \(F\), for each of the processes. First- and third-order dependencies are expected for the one- and three-photon processes, respectively. Saturation of the quantum dots is observed under UV excitation, with saturation intensity \(I_{\text {sat}}=3.3\pm 1.1\;{\mathrm {kW/cm}}^{2}\), found using a fit of the form:

$$\begin{aligned} F\propto \frac{s}{1+s} \quad s=I{\big /}I_{\text {sat}} \end{aligned}$$
(1)

By excluding data points greater than 50 % of the saturation intensity found for the UV excitation data, the index of the UV intensity dependence was found to be \(1.01\pm 0.15\). The index for the IR-induced fluorescence was \(3.08\pm 0.06\), confirming three-photon behavior. Saturation under three-photon excitation was not observed, but was estimated to occur at average intensities >1 MW/cm2 [12], which are not attainable without damage to the quantum dot sample. Third-harmonic intensity was also found to exhibit cubic behavior. As such, the fluorescence under IR excitation could occur either via a direct three-photon mechanism or via absorption of a third-harmonic photon. This experiment does not distinguish between the two mechanisms.

Fig. 4
figure 4

Fluorescence rate dependence of one- and three-photon excitation. a Intensity-dependent fluorescence measured under UV excitation shows a saturation intensity of 3.3 ± 1.1 kW/cm2 from the saturation fit, given by the dashed gray line. Excluding points above 50 % of the saturation intensity, and fitting a power law \(I^{n}\), gives \(n=1.01\pm 0.15\) (dashed blue line). b A power law fit of the intensity-dependent fluorescence measured under three-photon excitation is shown, with an index \(n=3.08\pm 0.06\) (dashed red line)

Full size image

5 Discussion

Our results show three-photon excitation and fluorescence in quantum dots, with the aid of ultrafast fiber lasers. With this comes the potential to improve the effectiveness of quantum dot-based fluorescent label schemes using low-cost hardware, including those used in biological systems. Telecom-band fiber lasers provide an economical source of laser hardware that has the potential to supplant more conventional laser hardware such as Ti:sapphire in many applications.

The experimental data for three-photon excitation show that there are no significant changes or shifts in the fluorescence spectrum compared to the linear excitation case, and the third-order dependence on excitation power shows that the quantum dot sample is indeed excited through a three-photon process. The process in the study by Horton et al. [13] could thus be replicated using quantum dots as a more favorable label agent [6, 8, 9].

6 Conclusion

We provide the first demonstration of three-photon excitation in quantum dots using a fiber laser operating in the telecommunications band. The experiment was conducted using a mode-locked Er-doped fiber laser producing 116-fs pulses with a center wavelength of 1,575 nm. We compare results from one- and three-photon excitation experiments to confirm that the three-photon spectrum is similar and to demonstrate the third-order intensity dependence of the process as an additional means of verification. We found that the fluorescence intensity scaled with input excitation power as expected for a three-photon process. Three-photon excitation in quantum dots using ultrafast fiber lasers provides an important step toward more effective imaging of fluorescent labels, particularly in biological applications, by minimizing scattering of the excitation light and increasing the possible penetration depth.