1. Introduction

Real-time on-chip measurement of optical refractive indices with high precision is of great interest in lab-on-a-chip devices and micro-electro-mechanical systems (MEMS) for ultra-sensitive detection of fluid flow in bio-sensing [1–3], chemical interaction monitoring [4,5], and environmental monitoring [6]. Several optical technologies have been reported for this purpose based on optical interferometry [7], surface plasmon resonance [8], Fabry-Perot cavity [9], whispering-gallery-mode micro-resonators [10–12], and fiber Bragg gratings [13]. Most of these techniques, however, require bulky hardware, precise optical alignment, or complicated manufacturing techniques [14,15], which are still very challenging. For wider-spread application of high-precision optical interferometric metrology, the optical systems should be miniaturized and easily integrated into the microsystems. Several methods have been investigated to remedy this problem, by introducing optical waveguides with external light coupling [16], embedding the optical fibers in polymeric devices [17,18], and integrating localized LED sources [19]. However, there are practical limitations associated with these techniques. The waveguide coupling suffers from low manufacturing reproducibility and sample leakage from the fluid channels [15,20]. In the case of fiber coupling, the alignment problems pose a major concern [21,22]. For the integrated localized sources, the electrodes tend to degrade over time, so are not reliable in the long term [15]. Therefore, there exist strong needs for a novel method that can facilitate the easier integration of coherent light sources into the micro-scale devices.

Optical micro- and nano- structures are manufactured through mask-based photo-lithography [23], mask-less electron-beam lithography [23], and focused ion beam processes [20]. Since these are subtractive manufacturing processes; repetitive coating, patterning and etching processes are necessary. Photolithography supports mass production well but as pre-patterned masks are required, it is not well-suited for complex 3D structures. The electron- and ion-beam lithography provide higher patterning resolution without masks but with a lower manufacturing throughput. Two-photon-polymerization (TPP) is based on the concept of additive manufacturing (so-called 3D printing), so arbitrarily shaped 3D polymer structures can be fabricated with a high design flexibility in a reasonably short build time [22,24–27]. Furthermore, because TPP is based on a nonlinear optical two-photon process, the printing resolution can overcome the traditional optical diffraction limit; therefore various novel micro- structures have been demonstrated in the past decade [24]. In addition, the photo-polymers such as SU-8 are well verified for their high optical quality and low propagation loss, so they are well-suited for refractive and diffractive optical structures [22,24–30].

Optical harmonic generation is a nonlinear optical process in which new frequencies that are integer multiples of the input laser frequency are generated, which can be used to generate new partially coherent light sources [31–34]. Traditional linear optical interferometry based on highly coherent laser beams suffers from coherent artifacts, such as coherent speckles, parasitic interference fringes among the multiple-reflected beams and other noise sources [35,36]. Nonlinear interferometry based on optical harmonic generation could confine the coherent volume and control the coherent artifacts, which have not been demonstrated much to date. Third-harmonic generation (THG) is an electric dipole allowed process that can occur in all materials, including materials with inversion symmetry. Although THG is usually a weak process, it becomes relatively stronger at the material interfaces [37–39]. Therefore, the surface THG from either crystals or dielectric materials can be used as novel micro-sized light sources [33,34,38,40–45].

In this article, we demonstrate a two-photon-polymerized micro-sized third-harmonic optical interferometer for the real-time, on-chip monitoring of microfluidic refractive indices. Lab-on-a-chip devices require compact coherent light sources for high-precision refractive index (RI) measurements within a small volume near to target samples. For this purpose, two micro-sized parallel cuboids were additively manufactured via TPP using 100-fs near-infrared femtosecond laser pulses; these structures worked as the nonlinear optical media generating coherent third-harmonic photons with high spectral purity. A simple vertical pumping via optical microscope enabled us to detect the resulting third-harmonic optical interferogram without complex light in- and out-coupling or precision alignment. The fundamental pump beam was filtered out, thus background-noise-free third-harmonic interferogram at the target volume was well extracted out. In the microfluidic measurement, a linearly polarized incident beam was used to generate the TH interferogram. A detection sensitivity of 0.22 (μm)⁻¹ (psi)⁻¹ was achieved in the argon gas flow detection. The real-time RI measurement was demonstrated up to 25 frames-per-second, which is limited by the camera frame rate, which can be easily improved to hundreds of kHz by introducing a higher speed camera.

2. Optical layout and sample fabrication

The micro-sized SU-8 structures are fabricated by the mask-free TPP and directly integrated into microfluidic systems as shown in Fig. 1(a). A simple vertical pumping of these printed structures results in highly-coherent compact third-harmonic (TH) light sources within a small volume near to target samples for real-time RI monitoring. Figure 1(a) right inset shows how the fabricated micro-sized TH optical interferometer can be easily integrated to an optofluidic device. The RI variations of target fluids can be analyzed utilizing the related spatial frequency changes in the TH interferogram.

 

Fig. 1 THG from 3D-printed SU-8 micro-structures. (a) Illustration of micro-optofluidic refractive index sensing with TH interferometry. (b) Optical and (c) SEM images of the TPP structures; the scale bar is 20 μm. (d) Optical layout of the TH analysis system. Abbreviations: HWP: half-wave plate, QWP: quarter-wave plate, LPF: long-pass filter, SPF: short-pass filter, BS: beam splitter, M: mirror, F-OBJ: focusing objective, I-OBJ: imaging objective, L: lens, CMOS: compact complementary metal oxide semiconductor camera and EMCCD: electron-multiplying charge-coupled device.

Download Full Size | PDF

To fabricate the micro-sized cuboids, a thin film of SU-8 (Microchem) spread on a glass substrate is firstly pre-baked at 65 °C for one minute and then at 95 °C for 25 minutes. After pre-baking, the SU-8 film is slowly cooled down to the ambient temperature (25 °C) over 10 minutes. The TPP structures are 3D-printed by scanning the near-infrared femtosecond laser’s focal spot (Toptica FemtoFiber pro NIR, 780 nm, 100 fs, 80 MHz) inside the SU-8 film along pre-programmed paths with a scanning speed of 20 μm⋅s⁻¹. After laser scanning, the sample is post-baked at 65 °C for one minute and at 95 °C for 5 minutes. The sample is cooled down to the ambient temperature and then immersed into the developing solution (Microchem) for 10 minutes to remove the unreacted photoresist. The sample is finally washed with isopropyl alcohol and dried inside a fume hood. The optical and scanning electron microscope (SEM) images of the printed SU-8 cuboids are shown in Figs. 1(b) and 1(c).

Figure 1(d) shows the experimental setup used for characterizing the TH output from the 3D-printed TPP. A wavelength-tunable supercontinuum fiber femtosecond laser (Toptica FemtoFiber pro SCIR, 980-2200 nm, 80 MHz) is used as the excitation light source. For real-time on-chip RI measurement, a slightly defocused beam should be illuminated over the edges of the two parallel cuboids; for efficient THG, a high peak power is required which is realized with a high average power of 170 mW and sub-100-fs short pulse duration at 1545 nm center wavelength. A half-wave plate (Thorlabs WPH10M-1550) and a quarter-wave plate (Thorlabs WPQ10E-1550) are employed to control the ellipticity of the incident laser’s polarization state. An 800 nm long-pass filter (Thorlabs FELH0800) is placed in front of the focusing objective lens to remove any residual short wavelength light present in the excitation beam. The beam is focused to a small spot of 7.6 μm diameter (FWHM: full-width-half-maximum) on to the sample using a 40 × objective lens (Newport LI-40X) with a numerical aperture (NA) of 0.65. A 20 × objective lens (Olympus LUCPLFLN 20X) of 0.45 NA is used to collect the nonlinear optical harmonics generated from the sample. A 750 nm short-pass filter (Thorlabs FES0750) is adapted to remove the excitation wavelength and the collected light is directed to a broadband spectrometer (Andor Shamrock 193i) connected to a highly-sensitive electron-multiplying charge-coupled device (EMCCD, Andor iXon Ultra) for spectral analysis. The measurement system was calibrated using a reference silicon wafer (non-doped, <100>, 280 μm in thickness) of known THG conversion efficiency. THG conversion efficiency at the printed structures was measured based on the EMCCD’s photon counts calibrated at this reference sample.

3. Results and discussion

3.1. THG characterization with elliptical polarization states and different interfaces

Figures 2(a) and 2(b) show how the ellipticity dependence of THG at different interfaces are characterized. A two-photon polymerized SU-8 cylinder of 40 μm diameter and 60 μm height is used for this characterization. The ellipticity of the input polarization state is controlled by rotating the quarter-wave plate while keeping the half-wave plate fixed, as shown in Fig. 2(a). THG from different interfaces are selectively excited by scanning the focal volume along the depth direction, on the air-glass, the glass-SU-8, and the SU-8-air interfaces, as shown in Fig. 2(b). The TH spectrum obtained from the SU-8-air interface is shown in Fig. 2(c). During these measurements, the 750 nm short-pass filter is removed from the beam path so that all the wavelengths of the collected beam are allowed to fall on the spectrometer for the initial characterization. It is worth to note that no second harmonic generation (SHG) is detected at half the excitation wavelength of 794 nm. This is because all the materials in the sample (namely air, glass, and SU-8) are isotropic and SHG is forbidden in materials with inversion symmetry. However, THG, an electric dipole allowed process, occurs in all materials including materials with inversion symmetry [37]. The central wavelength of the generated TH is 551.4 nm which is at one-third of the pumping wavelength. Detailed TH spectrum is measured with a broadband spectrometer coupled to an EMCCD. The harmonic generation nature is double-confirmed from the emission spectral bandwidth (0.005 eV) normalized by the harmonic order [38,46]; the optical harmonic generation provides a similar spectral bandwidth compared to the original bandwidth of the fundamental laser, which is 0.008 eV in our case, as shown in Fig. 2(d). For measuring the polarization ellipticity dependence, the quarter-wave plate (QWP) is rotated with a step size of 5 degrees from the initial position, which is set as 0 degree such that the fast axis of the QWP is parallel to the fast axis of the half-wave plate (HWP). The TH spectrum generated at each QWP rotation step is recorded by the spectrometer. The TH intensity decreased to 0 when the rotation angle is either 45 degree or 135 degree, where the incident laser beam is circularly polarized. The TH is at maximum when the incident laser is linearly polarized and the rotation angle of QWP is 0, 90 or 180 degrees (Figs. 2(e)–2(g)). It has been reported that THG vanishes when an isotropic media is excited with circularly polarized light, even at interfaces [47]. The polarization components of the THG in the case of an isotropic medium can be written as [48],

 

Fig. 2 THG dependence on input polarization ellipticities and focused target interfaces. (a) System schematic for measuring THG with different excitation ellipticities. (b) System schematic for measuring TH generated at different interfaces. (c) Optical spectrum of the generated TH. (d) Normalized TH bandwidth compared with that of the fundamental beam. (e) to (g): Normalized peak intensity spectra from air-glass, glass-SU-8, and SU-8-air interfaces with different polarization ellipticities of the incident beam. (h) Comparison of the TH intensities at different interfaces with different ellipticities of the incident beam. Abbreviations: HWP: half-wave plate, QWP: quarter-wave plate, LPF: long-pass filter, OBJ: objective.

Download Full Size | PDF

3.2. Linear polarization dependence and excitation position dependence of THG

For measuring the linear polarization dependence of THG, the QWP is removed and the HWP is rotated gradually with a step size of 5 degrees, as shown in Fig. 3(a). TH generated from all interfaces are independent of the polarization direction of the excitation beam, as shown in Figs. 3(b)–3(d). Since all the materials in the sample (namely SU-8, glass, air) are isotropic, there is no directional preference for THG. To find the excitation position dependence of the THG, the focused laser beam having a fixed linear polarization state is scanned through the polymerized SU-8 structure along the z-axis with a scanning step size of 10 μm. Figure 3(e) shows the TH intensity map in the sample at various excitation depths. The TH intensity is stronger at the interfaces and decreased to 0 within the bulk volumetric part of SU-8 and the glass substrate. The TH intensity reached its maximum when the laser beam is focused on the SU-8-air interface. For a tightly focused beam, the Gouy phase shift creates a phase difference of π between the THs generated before and after the focus and hence they destructively interfere, resulting in a low TH intensity in the bulk [38,41]. Since the symmetry along the optical axis is broken at the interface, the Gouy phase shift no longer results in total destructive interference at the interfaces. The TH intensity at the interface between two media scales approximately as ${\left({\alpha }_1-{\alpha }_2\right)}^2$, where $\alpha ={\chi }^{(3)}/n\text{Δ}n$, ${\chi }^{(3)}$ is the third-order susceptibility, n is the refractive index and $\text{Δ}n$ is the dispersion [49]. Since the third-order susceptibility difference at the SU-8-air interface is larger than that of air-glass and glass-SU-8 interfaces, the highest TH intensity is observed from the SU-8-air interface [50]. The strong surface TH from the SU-8-air interface of carefully designed SU-8 structures can be utilized as flexible coherent light sources to generate interferogram inside microfluidic channels, which can be used for real-time monitoring of the fluid’s refractive index.

 

Fig. 3 Linear polarization dependence and excitation position dependence of THG. (a) System schematic for measuring THG with different linear polarization states. (b) to (d): Normalized peak intensity spectra from air-glass, glass-SU-8, and SU-8-air interfaces with different input linear polarization states. (e) TH intensity distribution at different excitation depths along the sample. Abbreviations: HWP: half-wave plate, LPF: long-pass filter, OBJ: objective.

Download Full Size | PDF

3.3. Coherent TH interferometry using 3D-printed cuboids as light sources

Figure 4(a) shows a schematic illustration of the coherent micro-sized TH interferometry based on two spherical THG beams emitted from the TPP micro-cuboids. Two SU-8 cuboids with identical dimensions were fabricated; the width is 10 μm, the length is 40 μm, the height is 25 μm and the spacing between the cuboids is 8 μm. To excite both cuboids simultaneously, a slightly defocused incident pump beam is used with a beam waist of 28 μm. THs generated at the SU-8-air interface of the cuboids act as two coherent light sources and create interference patterns during propagation. Figure 4(b) shows the sample interference patterns recorded at different positions along the propagation. The contour map was generated using OriginPro from the intensity profiles recorded from 0 μm to 50 μm in steps 5 μm.

 

Fig. 4 Coherent TH interferometry using 3D-printed micro-cuboids. (a) Schematic illustration of the Young’s double slit interferometry created by the two coherent TH sources. (b) Normalized intensity map along propagation direction. (c) Spectrum of the incident laser beam. (d) Spectrum of the TH generated at the SU-8-air interface of the cuboids. Inset shows the optical image of the TH sources. (e) Spatial intensity distribution of the TH sources along the width of the cuboids. (f) Intensity distribution of the interference pattern imaged at 50 μm from the TH sources. (g) Spatial frequency distribution of the imaged interference pattern. (h) Zoomed-in view of the first spatial frequency peak. The inset shows a second order polynomial fitting to the peak. All scale bars shown are 10 μm.

Download Full Size | PDF

Figure 4(c) shows the spectrum of the incident pump beam. TH generated from the cuboids has a central wavelength of 533 nm and a spectral bandwidth of 21.85 nm, as shown in Fig. 4(d). The spatial intensity profile of the THG emission from both cuboids had high quality Gaussian shape with an R² value of 0.991, as shown in Fig. 4(e). The THG interferograms are imaged at 50 μm away from the TH sources and are analyzed using the Young’s double slit interference theory given by [7]:

3.4. Real-time on-chip monitoring of fluidic refractive index changes

The spatial frequency of the interferogram created by the coherent TH sources can be used for real-time monitoring of RI changes of fluids in microfluidic devices. A correction factor is required when applying the Young’s double slit equation (Eq. (3)) to account for the microfluidic system’s characteristics to the interferogram periods; however the spatial frequency still remains linear to the RI of the fluids. The TPP SU-8 micro-interferometer in Fig. 4(a) is incorporated into a microfluidic device to monitor the RI changes created by argon (Ar) gas flow through the device. The Ar gas pressure is controlled by a precision flow controller (Alicat Scientific MC-200), from 14.5 to 15.0 psi with 0.1 psi steps while the temperature of Ar is maintained at 28.1 °C. Figure 5(a) shows the TH interferograms recorded under different Ar pressures. The imaging position is set at a distance of 50 μm from the TH sources. When Ar pressure is controlled from 14.5 to 15.0 psi, the spatial frequency of the interferogram is found to increase from 0.495 to 0.605 μm⁻¹, as shown in Figs. 5(b) and 5(c). According to Gladstone-Dale model, the refractive index change of a gas (∆n) is linear to the change of gas pressure (∆P), given by [11]:

 

Fig. 5 Real-time monitoring of the refractive index changes of Ar gas under different pressures. (a) Recorded interferograms under different Ar pressures imaged at a distance of 50 μm from the TH sources. The scale bar is 10 μm. (b) Normalized spatial frequency profiles of the interference patterns under different argon pressures. (c) Polynomial fitting of the data points near peaks in Fig. 5(b). (d) Pressure dependence of the spatial frequency of interferogram.

Download Full Size | PDF

In this scheme, the surface quality of the printed structures should be precisely controlled and high peak power femtosecond laser pulses should be utilized for the TH generation. However, it must be noted that the use of infrared (IR) pump beam for generating the localized TH provides new possibilities. For example, the IR pump beam can pass through substrates that absorb visible light (such as silicon) and then generate localized visible light sources at the target location, which enables this scheme’s application into silicon waveguides and semiconductor characterizations [42]. This has been impossible with any other conventional light coupling schemes. The depth selectivity, suppressed background-noise, and new applicability in silicon wafers are the key strengths of the proposed scheme.

4. Conclusion

To conclude, we have thoroughly characterized TH excited at different interfaces of two-photon-polymerized SU-8 structures with different incident powers, input polarization states, and excitation depths. The highest conversion efficiency was observed at the SU-8-air interface (2.6 × 10⁻⁷) when excited with linearly polarized light, which is 4.93 times higher than that at SU-8-glass interface, and 12.17 times higher than that at glass-air interface. Due to the isotropic sample nature, the generated TH has no dependence on the linear polarization direction of the pump beam. Coherent THG from two printed cuboids are employed for the development of a micro-interferometer which can be easily integrated to microfluidic devices for the real-time RI monitoring. The real-time monitoring of Ar gas pressure is demonstrated in a microfluidic chamber with a high sensitivity of 0.22 (μm)⁻¹ (psi)⁻¹. The creation of the micro-interferometer via two-photon-polymerization offers high design flexibility; a simple vertical pumping of the polymerized structures for constructing the coherent interferograms eliminates the need for complex alignment or precise light coupling. Furthermore, background-free TH sources can be easily realized by removing the incident laser wavelength with band-control filters. Therefore, our scheme for realizing coherent TH micro-interferometer will offer versatile RI measurement solutions for integrated on-chip photonic applications.