Fig. 2. The optical setup and control scheme of the femtosecond laser imaging and microfabrication system.
2.3. Designing 3D freeform structures
In addition to nonlinear optical imaging capabilities, CAD software such as AutoCAD, Pro/E, and Solidworks can be used to design 3D structures for microfabrication. To transform 3D structures into 2D processing patterns, transformation programs such as Rhino and Materialse Magics can be adopted to convert the 3D structures into sequential 2D DXF files [31]. The 2D DXF files are then converted into bitmap files and downloaded into the FPGA module as laser processing commands. With the use of 3D structure design, we were able to create the desired structures.
3. Experimental results and discussions
With the fabrication solution consisting of acrylamide/bis-acrylamide, RB, TEA, DMSO, and AuNRs, we were able to improve the polymerization efficiency. More than 1.0 mM of the RB was required to provide adequate photoinitiation processing, while the DMSO surfactant is for the uniform dissolution of TEA. The AuNRs with sufficient CTAB can be dissolved in the fabrication solution. In order to implement the multiphoton fabrication of 3D polymer microstructures without femtosecond laser damage to the AuNRs, it is important to adopt a fabrication laser power as low as possible with a wavelength appropriate for the RB TPA, but not for the AuNR absorption. After multiphoton fabrication, the selectivity of AuNRs with different aspect-ratios in different locations can be achieved by photothermal reshaping. Moreover, a higher laser power, greater than the threshold of the AuNR damage and at the resonance wavelength of their longitudinal plasmon, was utilized to reshape the AuNRs into Au nanospheres. As a result, the existence of the AuNRs in designated positions of the fabricated microstructures can be achieved.
3.1. Wavelength selection in femtosecond laser microfabrication
In TPP processing, we can improve the polymerization efficiency of the acrylamide/bis-acrylamide monomer by adopting the laser wavelength at the maximum TPA of the photoinitiator, RB. Upon excitation, the time-averaged TPF photon count (F) of a fluorescence species is proportional to the cross section (δ) of TPA and can be given as [32]
where η2 is the quantum efficiency of TPF, φ the fluorescence collection efficiency of the detection system, C the concentration of the photoinitiator, gp the dimensionless quantity for the degree of the second-order temporal coherence, f the pulse repetition rate, τ the excitation pulse width at full-width at half maximum, n the refractive index of the measurement medium, P the average incident power, and λ the excitation wavelength. The excitation pulse widths can be maintained at different excitation wavelengths after the SF-10 prism pair compensation. The TPA can be represented by δ × η2
2 F
.
δη ∝λτ
, and based on the measured TPF photon count, the measured excitation pulse width, and the excitation wavelength. Hence, the TPA (probability) can be expressed as
(2) In TPA spectrum measurement experiment, the TPF photon counts were collected by the PMTs via the SPC module at the x galvanometer scanner rate of 20 kHz and the average excitation power of 10.0 mW. The excitation spectral range was selected from 710 to 830 nm and the pulse widths at different wavelengths after the objective were monitored by an in-lab constructed autocorrelator. The relative TPA spectrum of the RB (2.0 mM) in DI water as a function of the excitation wavelength is shown in Fig. 3. It was found that within the excitation wavelengths we examined, the excitation wavelength corresponding to the maximum value of the relative TPA of the RB was between 710 and 720 nm. Therefore, a fabrication laser wavelength of around 720 nm was adopted. Moreover, in order to implement the multiphoton fabrication of 3D polymer microstructures with AuNRs, the wavelengths of the two plasmon resonances of the adopted AuNRs should differ significantly from the fabrication wavelength of 720 nm. As shown in Fig. 1(a), the AuNRs with an aspect ratio of approximately 3.6 exhibits two plasmon resonances with transverse plasmon at around 520 nm and longitudinal plasmon at 780 nm in the fabrication solution. However, the AuNRs with a longitudinal plasmon wavelength longer than 780 nm are also good candidates for the 3D multiphoton fabrication process with the RB.
7000 720 740 760 780 800 820
Fig. 3. TPA spectrum of RB as function of excitation wavelength.
3.2. Selective AuNR reshaping by femtosecond laser
In order to implement multiphoton fabrication of 3D polymer microstructures, the power of the 100 fs femtosecond laser at the repetition rate of 80 MHz must be sufficient to support the MPE photochemistry process. According to our experience, the use of NA 1.3 objective and the x-galvanometer scan rate of 1 kHz, the laser power at the TPA wavelength of the photoinitiator must be controlled to within at least a few mW to implement the multiphoton fabrication in our solution. Furthermore, since AuNRs can be easily reshaped by utilizing the 100 fs laser at the resonance wavelength of the longitudinal plasmon of the AuNRs, the threshold power for completely melting AuNRs within a linearly polarized (along the longitudinal axis of the AuNRs) laser pulse was about 0.96 mW [33]. However, from the simulation results based on a two-temperature model, the threshold power for the 100 fs linearly polarized laser at the resonance wavelength is 0.049 mW [34].
When the polarization of the laser is turned from linear to circular, the threshold power can be decreased half.
Also, the orientation-specific damage for the long axis of AuNRs parallel to the direction of the linearly polarized laser can be avoided. From our results, the threshold power for AuNR melting using the circularly polarized, femtosecond laser at the resonance wavelength was 0.5 mW. This laser power would result in the minimal damage of AuNR for all of the orientations. Based on the AuNR absorption spectra in Fig. 1(a), the minimum AuNR damage power at the RB TPA wavelength of around 720 nm can be several times higher than 0.5 mW, the power used in processing the AuNRs at the longitudinal plasmon resonance wavelength of 780 nm.
In our experiments, a quarter wave plate (QWP) was inserted after the linear polarizer (Fig. 2). The fabrication laser power of 1.0 mW at the optimal fabrication wavelength of 720 nm was used (RB concentration 2.0 mM) [35]. Due to the generation of highly efficient TPL, the AuNRs can be attractive contrast agents for imaging 3D fabricated microstructures. Therefore, we examined the tomographic profile of the AuNRs filled fabricated microstructure by TPL imaging. Fig. 4(a) shows the TPL image of a fabricated 10 x 10 μm2 square polymer microstructure filled with AuNRs. In this study, TPL images were excited by the use of 0.1 mW, 100 fs laser at 780 nm and a scan rate of 20 kHz. The SEM zoom in image shows clear and intact AuNRs inside the polyacrylamide (Fig. 4(b)). These results indicate that there was no change to the morphology of the AuNRs after the femtosecond laser fabrication process, even when 1.0 mW fabrication laser power was used. It is likely that most of the laser energy was dissipated into the RB for
photopolymerization, not for AuNR absorption.
(a) (b)
Fig. 4. A 10 x 10 μm2 square polymer microstructure with AuNRs imaged with (a) TPL and (b) SEM.
The 780 nm AuNRs with different orientations in the 3D polymer microstructures can be selectively processed in designated positions via the photothermal reshaping mechanism. For this purpose, the ti-sa wavelength was tuned to 780 nm, the longitudinal plasmon resonance of the AuNRs, and circularly polarized laser was adopted to reshape the AuNRs in all of the orientations. To completely reshape the AuNRs, the laser power was increased to 5.0 mW. Fig. 5(a) shows the cross TPL image of the fabricated AuNR-doped microstructure (Fig. 4(a)) after photothermal reshaping to destroy the AuNRs outside of the cross pattern. The TPL image was acquired under the same imaging condition as in Fig. 4(a). The SEM zoom in image indicates that the AuNRs were completely reshaped into spherical AuNPs at 5.0 mW (Fig. 5(b)).
(a) (b)
Fig. 5. Cross pattern of AuNRs created by photothermal reshaping acquired with (a) TPL and (b) TEM.
3.3. 3D microfabrication
For 3D microfabrication, the sequential 2D bitmap files sliced from a 3D CAD model were downloaded into the FPGA to control the laser illumination via the AOM. Since TPP is confined to the focal volume, 3D freeform polymer solid structures can be developed. Unreacted solution was then washed out by water. Fig.
6(a) shows the TPL image of a hollow 3D microstructure by utilizing the fabrication and imaging conditions described in Sec. 3.2. The structure has a base area of 20 x 20 μm2 and a height of 5 μm. Finally, the diameter of the hole was 10 μm and the distance between two adjacent axial layers was 0.1 μm. Fig. 6(b) is the cross-sectional image of the base of the microfabricated structure of Fig. 6(a). The fabrication spatial
resolution of the 3D polyacrylamide microstructure we achieved is not as fine as ethoxylated trimethylolpropane triacrylate polymerization microstructures [31]. However, the water soluble acrylamide monomers were uniformly mixed with AuNR. Moreover, the polyacrylamide material is biocompatible, and therefore ready for bioapplications.
(a) (b)
Fig. 6. 3D TPP microstructure imaged by (a) 3D TPL (Insert: 2D bright-field image) and (b) 2D TPL image of the microstructure in (a) at the base.