1. Introduction

The contrast mechanism of third-harmonic generation microscopy (THG) relies on changes in the effective third-order nonlinear-optical susceptibility of materials placed in the focal volume of the microscope. Since all materials exhibit a non-vanishing third-order susceptibility, THG microscopy is applicable to any kind of sample. In addition, the THG signal vanishes inside an isotropic medium with normal (or positive) dispersion as a consequence of the Gouy phase shift as the beam passes through the focus [1]. This makes THG microscopy extremely sensitive to inhomogeneities such as interfaces between two materials [2–6]. Moreover, THG signals are also influenced by structures enclosed in the focal volume [7,8] and polarization of the incident electric field [9–15]. Indeed, THG in isotropic media is forbidden for circularly-polarized incident light [9], allowing the suppression of the unwanted background signals in biological [16,17] and synthetic materials [18,19].

Two-photon polymerization (TPP) is a versatile tool for fabricating sub-wavelength features that are localized in three dimensions [20–26]. Here, two-photon absorption is utilized to alter the chemical properties of a photosensitive polymer, which exhibits transparency in the visible and near-infrared ranges [20–22]. In addition, three-photon polymerization was employed to fabricate even smaller sub-wavelength features in a photoresist [27]. However, photopolymerized regions usually exhibit poor contrast against background when viewed under traditional bright field illumination. Although contrast-enhanced conventional optical microscopy techniques that rely on interference or phase can be used to visualize photoresist structures, the techniques are restricted by the sample thickness or require further modifications [28]. Furthermore, the traditional techniques used to visualize such microstructures lack three-dimensional imaging capabilities [29–31] and require physical sectioning or external contrast agents [27,32–37]. While near-field imaging [38] and vibrational spectroscopy [39–41] have been utilized to characterize photopolymerized microstructures, such techniques are not easily accessible due to the requirements of multiple laser sources and cumbersome instrumentation. Thus, there is an obvious need for simple and nondestructive characterization tools for photopolymerized structures. Although THG microscopy was recently reported in imaging of laser-written structures (or voids) in glass [42,43] and ablated polymers [44], the capability of the technique has not yet been reported for photopolymerized microstructures, where the image contrast between the polymerized structure and air or the unpolymerized photoresist is inherently poor.

In this Paper, we show that polarized THG microscopy can be used for noninvasive characterization of microstructures fabricated using photopolymerization. We demonstrate the applicability of the technique by imaging TPP-fabricated microstructures in a SU-8 film. We find that the contrast in the THG images is higher than obtained by conventional linear techniques. We also show that THG microscopy with circular polarization provides better contrast by removing the unwanted background signal. Finally, we apply the technique for simple, nondestructive and three-dimensional imaging of photopolymerized structures.

2. Material and methods

2.1. Sample preparation

Negative photoresist (SU-8 5, Microchem) was selected as the test material because this photoresist is widely used. The photoresist was spin-coated on a standard microscope cover slip, which was pre-cleaned with acetone and treated in piranha solution at 80°C for 20 minutes. The thickness of the photoresist film was found to be around 6-8 μm. The spin-coated samples were then pre-baked at 65°C for 1 minute and at 95°C for 3 minutes. The TPP-induced structures were post baked at 65°C for 1 minute and at 95°C for another minute. Finally, the structures were developed using the mr-Dev 600 developing agent (Microchem).

2.2. Experimental setups

A custom-built TPP direct laser writing setup was used to fabricate three-dimensional microstructures in the photoresist film [Fig. 1] [35]. The setup was based on an ultrafast pulsed laser (780 nm, 80 MHz, 290 fs), whose beam was spatially filtered, expanded and collimated before directing it towards a dichroic mirror. The dichroic mirror directs the writing beam towards an infinity-corrected microscope objective (50 × , numerical aperture of 0.75), which focuses the beam into a diffraction-limited spot size of 520 nm. However, the minimum spatial resolution of our setup can be much smaller than the spot size of the incident laser beam because the feature size of the photopolymerized structures is strongly influenced by the exposure conditions and photoresist properties. The spin-coated SU-8 film was mounted on a three-axis nanopositioner. The sample was scanned in the focus along pre-programmed paths in order to write line pair structures. The structures were written with a scanning speed of 10 μm/s and an average laser power of 30 mW, well below the damage threshold of ~50 mW verified independently. A brightfield imaging arm was added to view the region of interest in the sample.

 

Fig. 1 Schematic diagram of TPP setup. L: lens, P: pinhole, DM: dichroic mirror, C: camera, F: infrared blocking filter, O: objective, S: piezo-scanning stage, LED: light emitting diode.

Download Full Size | PDF

A custom-built transmission-mode nonlinear microscope equipped with a mode-locked femtosecond Nd:glass laser (1060 nm, 82 MHz, 200 fs) was used in all THG experiments [Fig. 2] [14,17]. The laser beam was spatially filtered, expanded and collimated before directing it to an infinity-corrected microscope objective (numerical aperture of 0.8). The objective was used to focus the beam onto the sample, which was mounted on a 3-axis nanopositioner. The transmitted THG signal was collected by another objective (numerical aperture of 0.5), discriminated by suitable infrared blocking and interference (353 nm) filters, and detected by a cooled photomultiplier tube. A pixel dwell time of 100 ms was used in constructing the nonlinear images. A bright field-imaging arm, which consists of a LED, camera and flip mirrors, was incorporated to view the region of interest in the sample. To achieve controllable linear and circular polarizations, motorized half- and quarter-wave plates were used. The performance of all the optics for different polarizations was carefully verified. Since there is no optical component between the quarter-wave plate and the objective, circular and linear polarizations of high polarization purity are always produced. In addition, the behavior of the THG signal at the interface of a reference sample, e.g., glass cover slip, and air is always checked as a function of quarter-wave plate angle. The THG signal at this interface is near the noise level of the detection system whenever the quarter-wave plate angle corresponds to circular polarization. All measurements were performed at room temperature. Image analysis was performed using MATLAB. For proof-of-principle demonstrations, we imaged TPP-written SU-8 microstructures at average excitation powers not exceeding 80 mW.

 

Fig. 2 Schematic diagram of THG microscope. L: lens, P: pinhole, H/QWP: half-/quarter-wave plate, FM: flip mirror, C: camera, O: objective, S: piezo-scanning stage, F: infrared blocking and THG filters, PMT: photomultiplier tube, LED: light emitting diode.

Download Full Size | PDF

3. Results

We then imaged the developed TPP-made microstructures using our THG microscope. We observe an appreciable background while imaging the photopolymerized structure with linear polarization [Figs. 3(a)–3(d)]. The background originates mainly from the air-glass interface [2]. To eliminate the unwanted background, we then illuminated the same area in the sample using circular polarization [9,16,17]. As depicted in Fig. 3(b), the background in the THG image was significantly suppressed for an equivalent input power [9]. Additionally, we observed that the contrast in the THG images increases with respect to the input power as expected [Fig. 3(c)]. Clearly, THG with circular polarization provides a higher image contrast than linear polarization [Fig. 3(d)]. In addition, we see a reversal of the corresponding THG signal levels of the microfabricated line pairs using linear and circular polarizations [Fig. 3(d)]. We attribute this to the inherent specificity of THG with circular polarization to three-dimensional anisotropy [9,16,17]. For example, small writing inaccuracies from sample tilt or possible swelling of the negative resist during development might have introduced the formation of interfaces that are oriented in three dimensions. We also note that we did not see any damage to the samples even at the high power level of 70 mW. These results strongly suggest that the technique can provide high contrast images, does not induce further modification of the material and is indeed nondestructive.

 

Fig. 3 THG images of TPP-made structures on SU-8 photoresist using (a) linear and (b,c) circular polarizations at an average input power of (a,b) 10 mW or (c) 70 mW. The used input polarization is shown by the green arrows. Scalebars = 5 μm. (d) THG signal line cuts across the structures using linear (blue) and circular (yellow orange) polarizations. The input power used for linear (circular) polarization is 10 (70) mW. The corresponding regions of interest in the THG images are marked with colored lines in (a) and (c). The maximum THG signal is about 250 (8000) counts per 100 ms for linear (circular) polarization.

Download Full Size | PDF

In order to further verify the THG nature of the nonlinear signals from the TPP-made structures, we investigated their power dependence. The THG signals for circular polarization from two regions in the structure [Fig. 3(b)] were measured and plotted as a function of the input average power. As expected for a three-photon (third-order) process, the plots follow the cubic dependence of the nonlinear signal with excitation power [Fig. 4]. This result confirms that the nonlinear signal obtained is due to THG from the photopolymerized structures.

 

Fig. 4 Power dependence of the THG signals obtained from the TPP-made structures. Circular polarization was used. The data were taken at the regions which are marked with colored dots in Fig. 3(b). Solid lines show the cubic fit of the THG signal.

Download Full Size | PDF

Next, we demonstrate the three-dimensional imaging of the TPP-made structures using THG. Figure 5(a)-5(e) are different optical sections from the TPP-made structures. The THG images were taken from different z planes (Δz = 1.5 μm), where the starting z plane (z₁) is located near the top of the line structures. These experiments were taken using circularly polarized light at an average input power of 60 mW. We account the variations in the THG images to the morphology and quality of the photopolymerized structures which vary in three dimensions. Figure 5(f) is the scanning electron micrograph of a similar developed TPP-made SU-8 structure which was fabricated using the same parameters. Here, the width of the individual line structure is around 800 nm and the height is about 6 μm. Our work implies that THG microscopy can be used to evaluate the three-dimensional uniformity of photopolymerized microstructures. Furthermore, THG imaging can be used to possibly identify the location of defects that are buried inside the polymerized structures or swelling features of polymerized structures that may occur during the development process, which are not accessible in traditional techniques. Finally, we also expect that the technique is general and applicable to a wide variety of light-sensitive polymers or even functionalized ones.

 

Fig. 5 (a-e) Depth-resolved THG images (Δz = 1.5 μm) of TPP-made structures using circular polarization at an average input power of 60 mW. The maximum THG signal detected is about 100000 counts/s. (f) Top-view scanning electron micrograph of a developed TPP-made microstructure. Scalebars = 5 μm.

Download Full Size | PDF

4. Conclusion

We have demonstrated the capability of polarized THG microscopy for the characterization of microstructures fabricated by photopolymerization. As THG requires only a single near-infrared laser wavelength, the technique provides a simple way to visualize photopolymerized structures. We also showed that the use of circular polarization in THG microscopy provides better contrast compared to that of linear polarization by removing the unwanted background. The technique is capable of imaging the quality of three-dimensional photopolymerized microstructures, making it suitable for noninvasive diagnostics of polymerized devices. In the future, it is anticipated that THG microscopy can be utilized for real time monitoring of photopolymerization, addressing also structures even before development.