1. Introduction

Coherent nonlinear optical (NLO) microscopies such as second-harmonic generation (SHG) and third-harmonic generation (THG) provide new techniques to investigate the properties of individual metal nano-objects. These imaging modalities exploit NLO signals that are very sensitive to the symmetry, i.e., shape, chirality, or orientation, of the nano-object and to the polarization of the incident field [1]. Furthermore, the enhancement of NLO signals near the vicinity of metal nano-objects makes coherent NLO microscopy a very simple and favorable all-optical tool for plasmonics research.

The majority of SHG and THG microscopy studies of metal nano-objects, however, have been restricted to the use of conventional polarization, namely, linear and circular. Linear polarizations (LP) largely favor the identification and in-plane orientation determination of metal nano-objects [2–5]. The use of circular polarizations (CP) in SHG imaging of metal nano-objects allows versatile probing of sample chirality [6,7]. On the other hand, THG imaging using CP facilitates the detection of metal nano-objects on glass by removing the unwanted background THG signal from the homogenous substrate [5].

Interestingly, there has been a renewed awareness of unconventional laser beams such as cylindrical vector beams (CVB) that exhibit full vector symmetry about the propagation axis [8,9]. When tightly focused by a high numerical aperture (NA) lens, azimuthal polarization (AP) results in a doughnut-like intensity distribution with purely transversal electric field components at the vicinity of the focus. On the other hand, radial polarization (RP) produces a strong longitudinal field component along the direction of propagation when tightly focused by a high NA lens [10,11]. As a consequence, RP is capable of producing a smaller focal spot size than LP. Furthermore, RP has been applied in orientation imaging that uses confocal fluorescence [12], confocal interference [13], photoluminescence [14], two-photon fluorescence [15], coherent anti-Stokes Raman scattering [16], and SHG [17–19]. Moreover, the value of RP in SHG imaging of metal nanoparticles has been demonstrated recently to show contrast for sub-wavelength features that are not easily detected by conventional polarizations or linear scattering [20]. In addition, theoretical works on THG with weakly focused vector beams [21] and tightly focused RP beams have been considered previously [22]. A numerical investigation of THG microscopy with tightly focused unconventional beams has been shown to give rise to polarization sensitivity since the third-order (material) polarization is strongly influenced by the 3D electric field components in the focal region [23]. Experimentally, however, there is still no attention given to THG microscopy using tightly focused RP or AP beams.

In this work, we investigate THG and SHG from individual metal nanocones using tightly focused CVBs. In addition, we show that the combination of THG imaging and CVBs can be used as structure-sensitive probes for 3D metal nano-object characterization.

2. Materials and methodology

2.1 Nanocone samples

Our sample consists of periodic arrays of silver nanocones with different particle-to-particle distances that were fabricated on a fused silica substrate using ultraviolet-nanoimprint lithography combined with electron-beam evaporation [24]. Shown in Fig. 1(a) is a scanning electron microscopy (SEM) image of the nanocones with a period of 500 nm. Typically, a fabricated nanocone has a base diameter of about 150 nm, a height of about 500 nm and a tip radius of about 15 nm. An ideal nanocone is a symmetrical 3D nano-object with an out-of-plane orientation that is determined by the direction of the cone axis. However, in practice some of the cones have defects that break the symmetry, a fact that we exploited in the experiments described below. As reported earlier, a strong field enhancement at the cone tip is only achieved when the incident polarization is along the cone axis [25]. Figures 1(b) and 1(c) show SEMs of a near-ideal nanocone and one with a blunt tip and defect at the base, respectively.

 

Fig. 1 (a) SEM image of a periodic array of silver nanocones on fused silica substrate. Close-up SEM images of (b) normal and (c) defective nanocones. The scale bars in (a) and (b,c) correspond to 500 and 100 nm, respectively.

Download Full Size | PDF

2.2 NLO microscopy with CVB setup

The schematic of our custom-built NLO microscope is shown in Fig. 2. A mode-locked femtosecond Nd:glass laser (wavelength 1060 nm, pulse length 200 fs, repetition rate 82 MHz) is used as an excitation source. After collimation, expansion and polarization-cleaning, the output beam is directed to an infinity-corrected and strain-free microscope objective (50 × , 0.8 NA). The objective is used to focus the beam onto the sample that is placed on a three-axis motorized piezo-scanner. The transmitted fundamental and NLO signals are collected by a second objective (0.5 NA). Appropriate dichroic (long-pass filter with a cutoff at 409 nm) and optical filters (infrared block, THG filter with 5 nm bandwidth centered at 355 nm and SHG filter at 530 nm with 16.5 nm bandwidth centered at 532 nm) are used to discriminate the THG and SHG signals from the fundamental frequency. Cooled photomultiplier tubes are used to detect the THG and SHG signals.

 

Fig. 2 Scheme of the NLO microscope with CVBs. LP: linear polarizer, RPC: radial polarization converter, L: lens, P: pinhole, FM: flip mirror, C: camera, D: dichroic mirror, O: objective, S: piezo-scanning stage, IF: interference filter, FF, fundamental wavelength filter: PMT: photomultiplier tube, WLS: white light source.

Download Full Size | PDF

In order to build an image, the detected THG and SHG signals are plotted as a function of transverse positions. In the experiments, a pixel dwell time of 50 ms and an averaging of 2 measurements per pixel are used. To achieve RP and AP with high polarization purity, the LP beam input is directed to a polarization converter (ARCoptix, S.A.) and a spatial filter. In all the imaging experiments, average power levels incident on the sample are kept below 2 mW, except in the power dependence measurements where up to 4 mW is used. The used power levels are found to be below the damage threshold of the samples as evidenced by the agreement of the measured NLO signal behavior from the expected power dependence plots and reproducibility of the scan images. A transmission bright field imaging arm, which is coupled by flip-mounted dichroic filters, is also used to view the sample region of interest. In our experiments, we performed NLO measurements on single nanocones in the 2-µm period array to minimize coupling effects from adjacent nanocones.

3. Results and discussions

Before imaging the nanocones using CVBs, the excitation power dependence of the NLO signals from an individual nanocone under a tightly focused RP was investigated using the same optical setup. As shown in Fig. 3, the illumination power dependence curves for THG and SHG follow the expected cubic and quadratic behaviors for three-photon and two-photon optical processes, respectively. The excellent agreement observed confirms that the measured NLO signals are due to THG and SHG from a single nanocone. In addition, we observed a higher yield in THG than SHG. We explain this by the fact that, being not constrained by material symmetry, constructive THG likely originates from a larger sample volume than SHG. More specifically, THG is limited by the penetration of the local field into the metal, whereas SHG is electric dipole-allowed only at the metal-dielectric interface.

 

Fig. 3 Illumination power dependence of the THG and SHG signal from a single nanocone excited by focused RP. The measured THG and SHG signals were fitted with cubic (solid line) and quadratic (broken line) power curves (p), respectively.

Download Full Size | PDF

We then tested our method for imaging individual nanocones. Shown in Fig. 4 (top row) are the THG images of a normal nanocone that is excited by focused AP and RP. As shown, a dark spot is observed at the exact location of the nanocone due to the absence of excitation fields at the center of a focused AP. At the location when a focused AP beam is directly centered with the exact position of the nanocone, the nanocone is not excited at all because the central part of a focused AP beam is dark. In fact, the nanocone is only illuminated when the center of the focused AP does not coincide with the exact position of the nanocone. At the off-centered positions of the focused AP, the resulting transverse field components can illuminate the nanocone as evidenced by the doughnut-shaped intensity distribution in Fig. 4. Moreover, the asymmetry in the THG intensity distribution, i.e., enhanced strips that surround the dark spot, is attributed to in-plane anisotropy of the nanocone base [20]. On the other hand, scanning with focused RP led to an intensity distribution that resembles a single and localized bright spot at the location of the nanocone. Such intensity distribution is attributed to the presence of an intense longitudinal field component at the focus of the RP beam that efficiently excites the nanocone tip [20,25]. Generally, this finding is in good agreement with previous coherent NLO microscopies with RP that detect out-of-plane orientations [16–20].

 

Fig. 4 THG and SHG images of normal (top row) and defective (bottom row) silver nanocones using focused AP and RP. The THG and SHG images are depicted in separately normalized color scales. Average power of 2 mW was used. Scale bar = 500 nm. The location of the nanocone is marked by a green circle. The numbers at the bottom right corner of each image represent the relative signal strengths found in each image. The solid and dashed lines (cyan) correspond to the reference lines of interest used in Fig. 5.

Download Full Size | PDF

Shown in Fig. 4 (bottom row) are the THG images of a defective nanocone under focused AP and RP. Again, we obtain a dark spot at the location of the nanocone due to lack of excitation fields at the center of a focused AP. Furthermore, an asymmetric doughnut-shaped intensity distribution that surrounds the dark spot was observed. We attribute such asymmetry to in-plane anisotropy at the nanocone base [20]. Interestingly, scanning with focused RP led to the formation of a dark spot at the location of the nanocone that is surrounded by an asymmetric doughnut-shaped intensity pattern. The lack of THG intensities at the exact location of the nanocone suggests that out-of-plane excitations did not produce a reasonable THG at the chosen excitation wavelength. We attribute this observation to defects of the nanocone such as the presence of a blunt or bent tip, and other overall shape or size variations, which produced weak local-field enhancements at the vicinity of the nanocone tip [20]. Furthermore, the obtained THG image under focused RP resembles a doughnut-shaped intensity pattern with enhanced side lobes that are oriented perpendicular to the corresponding THG image with focused AP. Such complementarity in the intensity images is expected for RP and AP imaging of nano-objects possessing in-plane anisotropy where transverse field components in the focal volume play a key role. Similarly, we associate the asymmetry of the side lobes to nano-sized defects at the base, which affect the NLO signal [20].

Lastly, we compare the THG images of the nanocones with the corresponding SHG images. As shown in Fig. 4, a good overall correlation is observed between the THG and SHG images of the nanocones under focused CVBs. These results suggest that both THG and SHG phenomena with CVBs are governed by the same overall features of the individual nanocones. This implies that both THG and SHG are mainly driven by the strong local fields at the fundamental frequency, which depend on the structural details of each individual nanocone. For the present samples, the local fields at the harmonic frequencies, on the other hand, play a lesser role.

In terms of spatial resolution, the transverse spatial resolution of RP could be easily defined by measuring the relevant point spread function of the generated longitudinal field component. If the optical probe strictly responds to the longitudinal field component of the excitation, e.g., a nanocone, the transverse spatial resolution can be known [10]. This result is exemplified by the images in Fig. 4 using RP. Since the THG signal scales with the 6th power of the field, THG microscopy with RP has a higher transverse spatial resolution than corresponding lower-order optical techniques. Upon inspection, we found that the line width of the THG hotspot using RP is up to 36% smaller than the corresponding SHG hotspot (Fig. 5). On the other hand, one could expect that the width of the intense parts of the doughnut-shaped intensity distribution can be used as a measure of the transverse spatial resolution of focused AP. In addition, since the width of the doughnut depends on the order of the optical process, it is anticipated that the transverse spatial resolution is better in THG with AP than in lower order excitation counterparts.

 

Fig. 5 Normalized intensity profiles formed by averaging five adjacent line profiles that were derived from the THG and SHG images of a nanocone under focused RP. The reference lines of interest are depicted in Fig. 4. The additional lines were taken at spacings of 1 and 2 pixels with respect to the reference line.

Download Full Size | PDF

We also expect that the use of CVBs, particularly focused RP, is more efficient in exciting nano-objects with out-of-plane anisotropies such as nanocones or nanopillars than conventional polarizations. Such nanostructures will not be efficiently excited using linear or circular polarizations since the generated longitudinal field at the focus is very weak. Essentially, the increased excitation efficiency provided by focused CVBs is highly advantageous for microscopy, where contrast is more important.

4. Conclusions

We have investigated THG and SHG from individual silver nanocones using tightly focused CVBs. We found a larger yield in THG with focused RP than SHG. Furthermore, we have shown the possibility of THG imaging of individual metal nanocones with focused CVBs. We found an excellent correlation between the imaging properties of the proposed technique and SHG with CVBs, suggesting that both THG and SHG phenomena with CVBs are governed by the same overall features of the individual nanocone. We also provided direct evidence that the transverse spatial resolution of THG imaging with CVBs, particularly RP, is better than SHG. Furthermore, we emphasize that THG microscopy with CVBs is inherently sensitive to the structural features of the nanocone. As an example of application, the technique can be used to image arrays of nanocones and assess the quality of the fabricated cones or similar structures. In general, our work extends the ever-growing utility of focused unconventional polarizations for coherent NLO imaging of nanostructures.