A thorough understanding of the optical and electronic properties of fluorescent materials is imperative for designing next-generation nanodevices, sensors, photovoltaics, and for investigating biological systems. The ability to visualize and intimately monitor ultrafast charge carrier dynamics in single nanostructures is a prerequisite to being able to predict and ultimately control charge and energy transport on the nanoscale. Performing studies of this nature requires techniques with both nanometer spatial and femtosecond temporal resolution. Time-resolved optical microscopy has proven capable of fulfilling both these criteria as well as providing transient spectroscopic capabilities [1]. The past decade has shown an upsurge in the development of several spatiotemporal imaging approaches [2–5]. Methods based on luminescence detection have been the most prevalent because of the ultrafast response and sensitivity to various material properties [6,7]. Many of these imaging techniques are equipped with sufficient spatial resolution for single particle measurements and even for localized excitation on single structures [3]. Most approaches, however, involve raster scanning or phase matching requirements, both of which increase measurement acquisition times and the possibility of artifacts. Additionally, most of these methods are limited by their temporal resolution capabilities and, as a result, femtosecond time resolution has eluded all except a few [4,8].

Ultrafast optical gating can be a versatile alternative to other luminescence detection methods because it can facilitate the combination of wide-field imaging and femtosecond time resolution. Optical Kerr-gating, which utilizes the Kerr electro-optical effect, has previously been primarily used for time-resolved fluorescence/luminescence spectroscopy with high temporal resolution as well as fast acquisition times [9–11]. The combination of an optical Kerr gate with a setup for wide-field luminescence imaging resulted in a technique that offered access to ultrafast dynamics in single nanoparticles in the visible spectral range [12,13]. The extension of femtosecond Kerr-gated imaging into the UV spectral region allows for the detection of emission from single wide-bandgap semiconductor nanoparticles and gives access to charge carrier dynamics in individual nanoparticles through a contactless method. This instrument is an important addition to the field of time-resolved optical microscopy, which has a paucity of techniques able to image transient UV luminescence from single particles [14].

The new setup is based on a previously published design for imaging in the visible spectral range [12,13,15]. The conversion of the instrument for operation in the UV involved optimizing signal collection within that range without compromising the high temporal and spatial resolution of the original setup. The core of the setup is an optical Kerr gate that is integrated into the optical train of a homebuilt microscope. It consists of a Kerr medium positioned between two crossed polarizers. The gate is based on the Kerr electro-optic effect where the induction of a transient birefrigence in a medium is initiated by an ultrafast laser pulse (gate pulse). This change in refractive index ($\mathrm{\Delta }n$) is accompanied by a phase shift ($\varphi$) in the orthogonally polarized components of the incident light leading to temporarily elliptically polarized light. The phase shift is given by

The source of ultrafast pulses used for the microscope is a Ti:sapphire oscillator (Coherent Mantis). The output of the oscillator is amplified in a 10 kHz amplifier (Coherent Legend Elite) producing 38 fs pulses at 800 nm. The amplified output is split to be used as a gating beam in the Kerr gate of the microscope and to pump a homebuilt noncollinear optical parametric amplifier (NOPA). From the NOPA, 25 fs excitation pulses at 580 nm were generated and then frequency doubled in a nonlinear crystal to allow excitation of the sample at 290 nm.

The schematic for the experimental setup is illustrated in Fig. 1 along with a photograph of the Kerr-gated microscope. Three identical, all-reflective Schwarzschild objectives (${\mathrm{SO}}_x$) (Davin Optronics x36 Reflecting Objective, $\mathrm{NA}=0.5$) are used to preserve the high temporal resolution and to minimize group velocity dispersion in the new setup. The excitation pulse from the NOPA is directed onto the sample (S) via a small mirror on the first objective (${\mathrm{SO}}_1$). Fluorescence emitted from the sample is collected by the first objective, passed through the first polarizer ($P_1$) (Laser Components colorPol UV 380 BC4), and then focused onto the Kerr medium (K) with the second objective (${\mathrm{SO}}_2$). Both the first polarizer (2 mm) and the Kerr medium were kept as thin as possible to minimize group velocity dispersion. The gate pulse is superimposed with the image of the sample luminescence with a second small mirror. The polarization of the linearly polarized gate pulse is rotated by 45° with respect to the first polarizer via a waveplate (WP). A third objective (${\mathrm{SO}}_3$) collects the recollimated light before it passes through a second polarizer ($P_2$). At this point the recollimated light can be either imaged directly on a CCD camera (Andor Ikon-M 912) or dispersed by a prism and then detected by the CCD. The spectrometer ($\sim 2\mathrm{nm}$ resolution) consists of a slit, BK7 prism, two curved mirrors, and a focusing lens. The thermoelectrically cooled CCD camera enables long exposure times, which compensates for the decrease in quantum efficiency ($\sim 55\%$) of the CCD camera in the UV ($<400\mathrm{nm}$) detection range. In addition, Eqs. (1) and (2) show that the Kerr effect is more efficient in the UV compared to the visible.

 

Fig. 1. (a) Schematic of the Kerr-gated optical microscope experimental setup. S, sample; ${\mathrm{SO}}_{1–3}$, Schwarzschild objectives; $P_{1–2}$ polarizers; K, Kerr medium: D, delay; WP, waveplate; F and BG, spectral filters. (b) Photograph of the optical column of the home-built microscope.

Download Full Size | PDF

Time-resolved measurements are obtained when the gate pulse and the signal overlap spatially and temporally inside the Kerr medium between the crossed polarizers ($t_0$). The time between the gate pulse and the signal is delayed by changing the path length of the gate pulse (D) with a PI M410 precision linear stage and PI Mercury C-863 DC motor controller. A BG37 short pass filter was used to block scattered gate light propagating along with the signal and a BG3 filter to block any residual excitation light not already blocked by the polarizers.

Since the CCD itself is not gated, any leakage of the crossed polarizers reduces the contrast between the gated signal and the background. The signal-to-noise ratio is predominantly limited by the gating efficiency of the Kerr gate and the noise of the leakage through the crossed polarizers. High contrast in the polarizers and the large Kerr response in the Kerr medium are thus essential for the Kerr-gated microscope.

To address the former, polarizers with a contrast ratio in the UV higher than 100,000:1 and transmittance better than 52% were utilized in the new setup. The temporal response of the gate depends on the Kerr constant of the medium and its thickness. Two different Kerr media, yttrium aluminum garnet (YAG) and fused silica (FS), were investigated in this study for measuring in the UV range. Figure 2 shows the measured instrument response function (IRF) for YAG and FS. A 360 nm excitation pulse from the NOPA, reflected off a transmission electron microscope grid placed at the sample position (S), was used for measuring the IRF. A 0.5 mm YAG window showed a slower response (217 fs) when compared with a FS window of the same thickness (146 fs). Decreasing the thickness of the FS Kerr medium from 0.5 mm to 0.15 mm reduced the temporal response to 85 fs. The increase in the temporal resolution between YAG and FS is expected because of the group velocity mismatch (GVM) between gate and signal pulse and, to a smaller extent, because of the difference in group velocity dispersion (GVD) in the materials in the UV. The GVM between 800 nm and 400 nm is $3.94\times {10}^{-10}\mathrm{s}/\mathrm{m}$ and $1.56\times {10}^{-10}\mathrm{s}/\mathrm{m}$ for YAG and FS, respectively. The GVM also explains the faster response in 0.15 mm FS when compared to 0.5 mm FS. YAG shows a higher GVD at 400 nm ($244{\mathrm{fs}}^2/\mathrm{mm}$) than FS ($98{\mathrm{fs}}^2/\mathrm{mm}$) which introduces the asymmetry in the IRF for YAG shown in Fig. 2. A fit employing a convolution with a $\text{Gaussian}•{\sin }^2$ function (cf. [16]) results in an almost identical width. A fit employing a convolution with a Gaussian of time-dependent width would be necessary to account for the GVD at 400 nm and would reproduce the asymmetry.

 

Fig. 2. Instrument response function for YAG 0.5 mm (217 fs) and fused silica 0.5 mm (146 fs) and 0.15 mm (85 fs) at 360 nm.

Download Full Size | PDF

YAG was chosen as the Kerr medium with which measurements were performed for this investigation because of the gating efficiency [27% peak transmission (PT)], compared with the 0.5 mm (4% PT) and 0.15 mm ($\sim 0.5\%$ PT) FS [16]. Changing Kerr media is easily done within the setup and serves as a simple method for tailoring the instrument to perform with either a higher temporal response, higher Kerr efficiency, or compromising between these parameters depending on the sample requirements.

ZnO was used to demonstrate the capabilities of the setup as well as the efficacy of the technique in general. ZnO is well suited for these measurements because it has a band edge emission at 380 nm, which is well within the detection range of the new microscope. Additionally, ZnO has been reported to show complex ultrafast dynamics in the UV [8,17]. Single ZnO nanowires were measured in this investigation. The nanowires were grown via a vapor-liquid-solid (VLS) method previously reported [18]. SEM measurements showed that the nanowires ranged from 200–300 nm in diameter and were approximately 50 μm in length. VLS grown nanowires possess a higher crystallinity compared to hydrothermal methods and have fewer visible emissions from defect states [19]. Nanowire samples were prepared by drop casting nanowires in an acetone suspension onto silicon wafers and allowing it to dry. All measurements were conducted under ambient conditions.

In the open gate configuration, polarizers parallel, steady state fluorescence images and spectra were obtained. Figures 3(a) and 3(b) are open gate, false color images of a single ZnO nanowire at low ($4.4\mathrm{mJ}/{\mathrm{cm}}^2$) and high ($13.2\mathrm{mJ}/{\mathrm{cm}}^2$) pump fluence. At low pump fluence, there is uniform photoluminescence observed along the full length of the nanowire. When the pump power is increased, emission near the end of the nanowire becomes localized and more intense than at other regions along the length of the wire. Enhanced emission from the end of the wire is indicative of amplified spontaneous emission (ASE) or lasing [20,21]. The power dependent open-gate spectra [Fig. 3(c)] also reflect this change with a fivefold increase in intensity above the estimated threshold fluence of $6.2\mathrm{mJ}/{\mathrm{cm}}^2$ [inset Fig. 3(c)]. A blue spectral shift and broadening in the spectral line width is also observed with an increase in pump intensity. The former has been attributed to a reduction of the refractive index with an increase in carrier density in the gain region [22]. The latter is characteristic of a sample heating effect [21]. Additional low energy modes appear in the spectra at higher excitation intensities. The occurrence of these redshifted modes is in conjunction with an apparent saturation or tapering off in intensity of blueshifted modes. This feature is consistent with mode competition occurring in single nanowires where nascent modes compete for optical gain while preexisting modes become quenched [22].

 

Fig. 3. Open-gate image of a single ZnO nanowire at (a) low fluence ($4.4\mathrm{mJ}/{\mathrm{cm}}^2$) and (b) high fluence ($13.2\mathrm{mJ}/{\mathrm{cm}}^2$). (c) Power dependent emission spectra of the same nanowire. The inset is the emission intensity dependence of the pump fluence.

Download Full Size | PDF

Figure 4 shows data collected from the same nanowire with the microscope operating in transient, closed-gate configuration. The evolution of the dynamics along the nanowire can be imaged at any time after excitation [Fig. 4(a)]. These images can be collected as a sequence of images or viewed as a movie with frames taken at designated femtosecond intervals (Visualization 1). The nanowire measured in this study shows a region of high emission intensity that spreads or propagates along the length, becoming a larger region of high intensity at longer time delays. Nanowires can act as nanoscale waveguides and the time-resolved images could be explained by guiding along the wire. Such detailed characteristics of the emission cannot be identified with ultrafast spectral data alone. The imaging capabilities of the system are hence a vital tool for a complete analysis of single nanostructures. Transients from regions of interest can also be extracted from the nanowire at different fluences [Fig. 4(b)]. This capability is of interest for nanowires that may exhibit particularly complex dynamics in different locations along their lengths. The transient traces extracted from the region demarcated by the red box [Fig. 4(a)] provide insight into the emission intensity profile of the emitting region with time. These traces not only facilitate the observation of ultrafast mechanisms in discrete locations along a wire, but can also be used to determine carrier relaxation times and charge carrier mobilities in single nanowires [23].

 

Fig. 4. (a) Transient image sequence of a single ZnO nanowire excited at $11.5\mathrm{mJ}/{\mathrm{cm}}^2$ (see Visualization 1). (b) Transient dynamics extracted from the region on the wire demarcated by the red box in (a) for excitation fluences $9.7\mathrm{mJ}/{\mathrm{cm}}^2$ and $11.5\mathrm{mJ}/{\mathrm{cm}}^2$.

Download Full Size | PDF

Time-resolved spectra (Fig. 5) are useful for observing spectral changes in emission with time within a single nanowire. Phenomena such as time-dependent bandgap renormalization can be readily detected with the setup in spectral mode. Figure 5(a) is the false color time-resolved spectrum of the luminescence at excitation fluence $16.4\mathrm{mJ}/{\mathrm{cm}}^2$. Spectral intensity shifts can be observed with increased delay time. Figure 5(b) shows slices extracted from the spectrum in Fig. 5(a) that provide more detail by showing the emergence and disappearance of emission peaks at a 1.0 ps time delay. A full analysis accompanied by further characterization of this sample and similar nanowires is, however, beyond the scope of this Letter, but will be the subject of future investigations.

 

Fig. 5. (a) False color scale transient emission spectrum of the same ZnO nanowire at $16.4\mathrm{mJ}/{\mathrm{cm}}^2$. (b) Slices of emission spectrum in (a) taken at increasing delay times.

Download Full Size | PDF

In summary, we have successfully demonstrated the conversion and operation of the femtosecond Kerr-gated fluorescence microscope in the ultraviolet. To the best of our knowledge this is the first ultrafast Kerr-gated imaging system with sub 90 fs temporal resolution capable of detection in the ultraviolet. YAG and FS Kerr medium were compared. YAG had the higher gating efficiency (27%), while FS displayed a faster temporal response. The UV emission of a single ZnO nanowire was analyzed by steady state and transient imaging with micron resolution and steady state and transient spectroscopy using the new microscope.