Yasutaka
Hanada
ab,
Koji
Sugioka
*a and
Katsumi
Midorikawa
a
aRIKEN-Advanced Science Institute, 2-1 Hirosawa, Wako, Saitama, Japan. E-mail: ksugioka@riken.jp; Fax: +81 48462 4682; Tel: +81 48467 9502
bHirosaki University, Graduate School of Science and Technology, 3 Bunkyo-cho, Hirosaki, Aomori, Japan
First published on 14th June 2012
The demand for increased sensitivity in the concentration analysis of biochemical liquids is a crucial issue in the development of lab on a chip and optofluidic devices. We propose a new design for optofluidic devices for performing highly sensitive biochemical liquid assays. This design consists of a microfluidic channel whose internal walls are coated with a polymer and an optical waveguide embedded in photostructurable glass. The microfluidic channel is first formed by three-dimensional femtosecond laser micromachining. The internal walls of the channel are then coated by the dipping method with a polymer that has a lower refractive index than water. Subsequently, the optical waveguide is integrated with the microfluidic channel. The polymer coating on the internal walls permits the probe light, which is introduced by the optical waveguide, to propagate along the inside of the microfluidic channel. This results in a sufficiently long interaction length between the probe light and a liquid sample in the channel and thus significantly improves the sensitivity of absorption measurements. Using the fabricated optofluidic chips, we analyzed protein in bovine serum albumin to concentrations down to 7.5 mM as well as 200 nM glucose-D.
However, in all these light detection methods, the interaction length of light with the sample is determined by the width of the capillary or microfluidic channel, which limits the detection sensitivity. The interaction length would be greatly increased if the probe light could propagate inside the channel. However, materials typically used for optofluidic chips such as glass and polydimethylsiloxane (PDMS) have higher refractive indices than those of liquid samples, which prohibits light propagation inside the channel. To overcome this problem, in this study, we develop a technique for coating the internal walls of a microfluidic channel with a polymer that has a lower refractive index than water. The fabricated optofluidic chips are used to detect low concentrations of proteins and glucose-D with a high sensitivity.
Fig. 1 Optofluidic chip design for liquid concentration measurement. The optofluidic chip is filled with liquid and a white light probe is coupled to the entrance facet of the waveguide. The light transmitted by the waveguide propagates inside the microfluidic channel and is then reflected at the tilted wall of the left reservoir, where the absorption spectra of liquids are measured. |
To realize highly sensitive analysis of liquid concentration, the embedded microfluidic channel was coated with a low refractive index polymer (Teflon AF, DuPont) whose refractive index (1.31) is lower than that (∼1.33) of water. Thus, the probe light for liquid analysis can propagate inside the microfluidic channel filled with liquid samples, resulting in a sufficiently long interaction length between the liquid and the incident probe light, thereby increasing the analysis sensitivity. This coating was performed by dipping the optofluidic chip in the liquid polymer for 5 s and removing it with a speed of 4 mm s−1. The viscosity of the liquid polymer is sufficiently low that the polymer filled the microfluidic channel by merely dipping. An optofluidic chip with its channel filled with the liquid polymer was then annealed by increasing the temperature to 200 °C at 20 °C min−1 and holding this temperature for 2 h. Since the liquid polymer has a high volatility even at room temperature, the final thickness of the solid polymer coating on the wall of the microfluidic channel was about 1.7 μm.
To efficiently couple the probe light into the microfluidic channel, an optical waveguide was integrated between the channel and the right edge of the optofluidic chip. To write the optical waveguide, a focused fs laser beam with a pulse energy of 0.3 μJ was scanned at 1.5 mm s−1 inside the glass to increase the refractive index in the laser-scanned regions after polymer coating. This waveguide enables the probe light to be efficiently introduced into the microfluidic channel by an objective lens, as shown in Fig. 1. The written line functions as a single-mode waveguide with a propagation loss of 0.5 dB cm−1 at a wavelength of 632 nm.16
A glucose assay kit from BioVision was used for the glucose-D concentration assay. This kit contains glucose standards, an assay buffer for dilution, and an enzyme mix for inducing color change that depends on the amount of glucose present in the liquid. Glucose-D samples with different concentrations were prepared by diluting the glucose standard using the assay buffer (0.2, 0.4, 0.8, 1.2, 2 μM). After preparation, incubation was performed for 30 min for this reaction.
Fig. 2 (a) Near-field pattern and (b) intensity distribution of 632 nm CW laser beam at end of the microfluidic channel. The polymer-coated microfluidic channel essentially functions as a single-mode waveguide at a wavelength of 632 nm. |
Fig. 2 reveals that the microfluidic channel essentially functions as a single-mode waveguide at a wavelength of 632 nm. The intensity profile has a full-width at half-maximum of approximately 200 μm in the x-direction (i.e., parallel with the chip surface) and 350 μm in the y-direction (i.e., perpendicular to the chip surface), which is comparable to the cross-section of the microfluidic channel.
To evaluate the propagation loss by the polymer-coated microfluidic channel, five optofluidic chips with different microfluidic channel lengths were prepared. Fig. 3 shows the optical loss as a function of the microfluidic channel length. Much care was taken to ensure that the coupling conditions were identical for all five microfluidic channels. To determine the optical loss, a 632 nm wavelength CW laser was coupled to the waveguide by a ×20 objective (NA: 0.46). The coupling loss was estimated to be ca. 0.15 dB by extrapolating the data to a distance of 0 mm. The coupling loss is caused by either a mismatch between the size of the beam and that of the waveguide or misalignment between the waveguide and the focused laser. The propagation loss in the microfluidic channel is estimated to be 0.55 dB mm−1 at a wavelength of 632 nm from the slope of the data in Fig. 3. The CW laser light may leak through the thin polymer into the glass by its evanescent wave, which would cause propagation loss. However, the penetration depth of the evanescent wave is smaller than the laser wavelength, while the polymer thickness is larger than the wavelength. In all the propagation loss measurements, we did not observe the laser beam leaking into the glass. Although the propagation loss obtained is much greater than that of optical waveguides written by fs lasers, it is still acceptable for optofluidic applications due to the reasonably short channel length (in the present case, the microfluidic channel is ca. 2 mm long; the length is inevitably determined by the size of the glass chip and the arrangement of the left reservoir and the fs laser-written optical waveguide).
Fig. 3 Optical loss as a function of microfluidic channel length. The coupling loss is estimated to be ca. 0.15 dB by extrapolating the data to a distance of 0 mm. The propagation loss is estimated to be 0.55 dB mm−1 at a wavelength of 632 nm. |
In the concentration analysis, the laser beam is reflected from the tilted wall in the left reservoir after propagating through the microfluidic channel. The reflectivity at a wavelength of 632 nm was measured to be 40%. This large propagation loss and low reflectivity may be due to the roughness of the coated polymer, which is shown by the atomic force microscopy (AFM) image in Fig. 4. To observe its surface morphology, the polymer was coated on a photostructurable glass surface by employing the same dipping procedure described above. Despite the fabricated microfluidic channel having a smooth surface after additional annealing,10 the coated polymer had a root mean square surface roughness of 56.8 nm. The propagation loss and the reflectivity could be improved by reducing the roughness of the coated polymer. Nevertheless, the fabricated optofluidic chips exhibited good performance in concentration analysis, as described below.
Fig. 4 AFM image of polymer coated on glass. To measure the surface roughness, a low refractive index polymer was coated on the photostructurable glass surface by dipping. The root mean square roughness was measured to be 56.8 nm. |
Fig. 5 Absorption measurement of BSA standards using the optofluidic chip. (a) Absorbance spectra of BSA standards with different protein concentrations. (b) Standard curve of OD for protein concentration was obtained by plotting the peak intensity in (a). A detection limit of 7.5 mM protein concentration was obtained using the optofluidic chip. |
After demonstrating the highly sensitive detection of protein concentrations, we compared the sensitivities of the fabricated optofluidic chip used in this study and one used in our previous study.6 In the previous study, a simple straight microfluidic channel was first fabricated in glass. We then formed straight optical waveguides that intersect the microfluidic channel at its center by scanning a tightly focused fs laser beam using the same technique described above. After fabrication, we measured the absorptions of the same BSA standards as those used above with different protein concentrations.
Fig. 6(a) shows absorbance spectra of BSA standards with different protein concentrations using the optofluidic chips fabricated in this and previous studies (the ellipses labeled 1 and 2 indicate spectra obtained by the optofluidic chips used in this study and the previous study, respectively). Fig. 6(b) shows enlarged spectra from ellipse 2 in Fig. 6(a). At high concentrations, both optofluidic chips can detect the BSA standards. However, the intensity obtained by the optofluidic chips used in this study is enhanced by a factor of ca. 14 relative to that obtained by the optofluidic chips used in the previous study. Furthermore, spectra for protein concentrations in the range 7.5–37.5 mM obtained by the optofluidic chip used in the previous study almost overlap (see Fig. 6(b)), making it impossible to distinguish absorbance spectra obtained at different concentrations. Thus, the detection sensitivity of the previous fabricated optofluidic chip is limited to a concentration of 75 mM. This low sensitivity is due to the short interaction length between the white light propagating across the protein solution in the microfluidic channel (width: 0.2 mm). In contrast, the present optofluidic chip has an interaction length of 2.0 mm (corresponding to the channel length). Its sensitivity could be further enhanced by increasing the channel length. However, a longer channel would increase the optical propagation loss. Therefore, it is also necessary to reduce the propagation loss.
Fig. 6 Comparison of absorption measurements of protein concentrations using the optofluidic chip fabricated in the present study and one fabricated in a previous study. (a) Absorbance spectra obtained by optofluidic chips fabricated in this and previous studies (ellipses 1 and 2 indicate spectra obtained from the optofluidic chips used in this and previous studies respectively). (b) Enlargements of spectra indicated by ellipse 2 in (a). |
To evaluate the performance of the optofluidic chip for various liquids, detection of glucose-D solutions with different concentrations was also performed using the same optofluidic chip as that used for protein concentration analysis. Fig. 7(a) shows absorbance spectra of glucose-D solutions with different concentrations mixed with an enzyme and Fig. 7(b) shows the standard OD curve for glucose-D obtained from Fig. 7(a). The optical absorption spectra exhibit a high absorbance peak at wavelengths of about 550 nm. This peak drops gradually as the concentration decreases. A glucose-D concentration as low as 200 nM was clearly detected, indicating that the detection limit was improved by a factor of over 104 relative to that obtained (5 mM) by the microfluidic chip integrated with an unbalanced Mach-Zehnder interferometer described in Ref. 7. The lowest concentration detected in the present study was limited by the concentration of glucose-D solution available. Therefore, it may be possible to detect samples with concentrations lower than 200 nM.
Fig. 7 Absorption spectra of glucose-D solutions obtained using the optofluidic chip. Absorbance spectra of (a) glucose-D with different concentrations. (b) Standard curve of glucose-D amount for different concentrations by plotting the peak intensity in (a). A glucose-D concentration down to 200 nM was clearly detected using the optofluidic chip. |
Footnote |
† Published as part of a themed issue on optofluidics. |
This journal is © The Royal Society of Chemistry 2012 |