16. Plastic Lab-on-Chip for the Optical Manipulation of Single Cells

As previously discussed, the main limitation of μIM technology is the low reconfigurability of the mould once its geometry is defined. In fact, any modification to the mould cavity is very expensive. Our approach is based on a fixed master mould and several exchangeable inserts, which can be easily replaced to produce devices with different functionalities. In this way, the complexity of the mould increases, but it allows a significant improvement in its versatility. The combination of different micro-manufacturing technologies has been evaluated in order to produce inserts that can be easily assembled in the mould and that can provide the microfeatures required by the LoC design.

Femtosecond laser ablation of steel has been used to produce the negative of the microfeatures required on the plastic devices. This technology allows to process reasonably large surfaces with small depth features and at high removal rate. In addition, it provides geometrical precision and surface roughness values that meet the requirements of the microfluidic application. To implement easy connection of the LoC with external tubing, complex 3D micro features with high aspect ratio have been produced with micro Electro Discharge machining (µEDM) technology.

In the following subsections the design of the LoC, the mould design and the tool insert manufacturing are described in detail.

The targeted device is an optical stretcher, consisting of two optical fibres, facing each another on the opposite sides of a microfluidic channel. The counter-propagating light beams delivered by the two fibres exert optical forces on the passing cells, trapping and deforming them. A microscopic imaging of the cell deformation for calibrated optical forces allows retrieving the mechanical compliance of each cell. The conceptual design of the device (Fig. 16.1) is made by two shells (19.5 mm length, 7.5 mm large and 0.7 mm thick) assembled together by femtosecond laser welding [28]. On the lower shell, a microchannel of 100 µm width and 90 µm depth with final reservoirs of 1 mm diameter is produced. V–grooves have been realized for the optical fibre housing with depth and width of 180 µm and 235 µm, respectively. Connectors on the upper shell have been conceived to reduce the number of external components involved in the assembly, thus enabling a direct connection of external capillaries to the chip.

The two polymeric shells were finally assembled by direct fs-laser welding. Although already successfully demonstrated for glasses, this kind of approach was never applied to transparent polymeric substrates.

Firstly, the combinations of process parameters producing melting of the laser focal volume at low pulse energies and high repetition rates were estimated by solving the heat conduction equations describing the pulsed-laser irradiation of a PMMA substrate [28]. Then, the modifications induced by focusing fs-laser pulses in the volume of the bulk material were experimentally investigated. The goal of this study was to identify an appropriate window of process parameters producing homogeneous and continuous melting, spatially localized in the focal region. This time the laser beam focus was achieved through a 0.3 numerical aperture lens and positioned 1.5 mm below the surface of a 3 mm-thick PMMA specimen. The beam was then moved along single and multiple parallel lines. Several tests were conducted by operating the laser source at 200 kHz, 500 kHz, 1 MHz, 5 MHz with average power ranging from 0.1 to 1 W. The translation speed was varied from 0.01 to 10 mm/s. Each test was repeated with 650 fs and 18 ps pulse durations in order to understand the influence of this parameter on the material modification.

The tests demonstrated that the peak power has a significant role in triggering changes of the material. Therefore, we scaled the pulse energies at the two pulse durations to have the same peak power. At both pulse durations, it was evident that heat accumulation is enhanced at higher repetition rates (>1 MHz) and low translation speeds (<1 mm/s), in agreement with the simulation results. We observed that better quality results are achieved with 650-fs pulses rather than with 18-ps pulses, therefore we chose the shorter pulses for performing this process.

Based on the results of these preliminary tests, a suitable set of laser parameters was selected producing homogeneous melting in the focal region. The laser source was operated at 5 MHz with pulse energy of 0.4 μJ. The two injection-moulded PMMA shells (Fig. 16.7) composing the optical stretcher were mechanically clamped to achieve an air gap of a few micrometres in the welding area. The laser beam was focused at the interface between the two plates and was moved with a translation speed of 0.1 mm/s. The laser-welding path surrounded the microfluidic network composed of the two reservoirs, the channel and the two V-grooves.

As a preliminary step, the micro-lenses were integrated onto a commercial PMMA microfluidic device (ChipShop GmbH) through forward pyro-printing [33]. Pyro-EHD printing is a novel technique to develop micro-lenses integrated on LoCs while guaranteeing high chemical compatibility, good optical characteristics and flexibility in the design of the realized devices [34].

Micro-lenses fabrication was performed using the printing configuration showed in Fig. 16.7a. A polydimethylsiloxane drop (PDMS Dow Corning Sylgard 184, 10:1 mixing ratio, base to curing agent Midland; η = 3900 cP, εp = 2.65 from datasheet), was deposited as a reservoir drop on the pyroelectric crystal and placed above the PMMA device. Once an external heating source is applied onto the crystal, the PDMS drop starts to deform assuming a conical shape, known as Taylor’s cone, and dispensing micro-nano and pico-drops [35]. Figure 16.7c shows pictures of a microfluidic device with PDMS micro-lenses pyro-ink-jet-printed on it. Micro-lenses could present different shapes and dimensions, as a function of the pyro-inkjet printing parameters. Once the lenses were deposited on the channel, the device was heated at 80 °C for 30 min, in order to cure the polymer.

A fluorosilane agent (Fluorolink S10, Solvay Solexis) was used to modify the substrate wettability to avoid excessive spreading of the PDMS lens onto the PMMA substrate. Moreover, the antireflection properties of the coating guarantee good optical properties of the final device. Furthermore, fluorosilane has good chemical resistance and endures abrasion. An O₂ plasma treatment (Femto System, Diener Electronic GmbH & Co. KG, Ebhausen, Germany) was performed on the microfluidic device to improve the adhesion between the fluorosilane agent and the PMMA substrate.

The µIM production of the two PMMA shells is based on the approach proposed in [36]. The process parameters defined with the simulation and used for the production of the components are reported in Table 16.1 where T_melt is the melt temperature, T_mould is the mould temperature, V_inj is the injection speed, P_hold is the packing pressure, t_hold is the packing time, t_cool is the cooling time and Run is the plunger injection run.

The µIM machine used for the device production is a DESMA Tec Formica Plast 1 K and the polymer Acyrex CM211 PMMA was selected as material for the device fabrication.

The adopted strategy to reduce mould roughness was to use a very low spark energy level during µEDM milling and to adjust the number of scanning loops during femtosecond laser ablation. With the selected parameters (reported in previous sections) we obtained microfluidic channels with a roughness (Ra of 0.59 µm) and a shape that match the designed values (see Fig. 16.8). The surface of the access holes on the upper shell presented a relatively low roughness (Ra of 0.37 µm) if compared with raw material. The capillary tube was easily inserted in the access hole providing a good housing adhesion when the fluid flew.

Finally, the V-grooves were realized orthogonal and coaxial to each other (Fig. 16.8) to allocate two standard optical fibres perfectly aligned and emitting two counter-propagating beams at a height of 20 µm with respect to the bottom of the microchannel. This height was chosen to maximize the number of cells intercepted by the optical trap. The obtained feature dimensions (see Fig. 16.8) are within 6% of the expected ones.

To perform optical imaging of the specimens inside the microfluidic channels, it is mandatory to have a channel bottom surface with optical quality. We developed a chemical post-treatment to smoothen the channel surface: the chip was exposed to Chloroform vapour (at 40 °C) for a few minutes and then annealed at low temperature (80 °C) for 20 min [37]. In this way, the roughness of the surface was reduced (see Fig. 16.9, showing the microchannel bottom before and after the treatment). The improved quality of the surface enabled accurate imaging of the elements flowing in the channel. In addition, this smoothening process is very simple and can be applied in parallel to several chips. Therefore, it relaxes the specifications on the quality of the micromachined sample allowing faster and cost-effective fabrication of the devices.

Firstly, static leakage tests were carried out. A dead-end reservoir, composed of two laser-bonded PMMA slides, one of them preliminarily laser-drilled, was connected to a microfluidic pump (MFCS-EZ, Fluigent) [28]. A blue-coloured liquid was injected at a pressure between 100 mbar and 1 bar with steps of 100 mbar, retaining each value for 3 min. The possible occurrence of fluid leakages was monitored through an optical microscope.

Further leakage tests were executed on a laser milled microchannel, which is the main core of any LoC. The microchannel was fabricated on the surface of a 1-mm-thick PMMA layer by direct femtosecond laser ablation, resulting in a square cross section of 80 μm side and a length of 1 mm. The channel connected two laser micro-machined reservoirs, each of them provided with a fs-laser drilled hole. This micro-machined layer was laser bonded to a standard PMMA slide all along the perimeter of the channel and of the reservoirs. A blue liquid was pumped in one reservoir, while the other end was hermetically closed. The leakage test was repeated several times on the same device to demonstrate that the laser bonded contour was able to withstand pressure over time [28].

The lenses were characterized interferometrically. In fact, an interferometric characterization based on Digital Holography (DH) microscopy was carried out to evaluate the optical aberrations of the microlenses and their focusing properties. The optical set-up was a classical Mach-Zehnder interferometer in transmission configuration [33]. Once the hologram was recorded, numerical propagation provided the object complex field in whatever plane along the optical axis z, from which the intensity and the phase distributions of the optical wave field transmitted by the sample could be extracted. A 2D fitting was applied to the recovered phase map by using a Zernike polynomial expansion to model and study the optical aberrations produced by the lens. The analysis of the first 10 orders of the linear combination of Zernike polynomials showed that, after the constant offset (P = 1), the main contribution to the development of the Zernike function was a defocus term (P = 5). This result was expected because of the spherical shape of the lens. Furthermore, a tilt along the y-axis (P = 3) was identified together with a coma aberration along the x-axis (P = 8). The other terms gave a negligible contribution [33]. After completing the optical characterization of the polymeric micro-lenses, their imaging capabilities were tested to show one of their possible applications on a LoC platform. Fibroblast cells were inserted to flow inside the microfluidic channel where the lenses were deposited and aligned. The images of the cells created by the on-chip lens were further imaged by an optical microscope [33]. Figure 16.10a, b show the images of the cells inside and outside the lens area. In particular, these are clearly in-focus only when they flow below the lens, while these are out-of-focus in all the other portions of the channel. In Fig. 16.10a it is also possible to notice the magnifying action operated by the lens, as the channel itself appears to be enlarged. Moreover, in order to proof the lens magnification, a test resolution target placed behind the chip is imaged. It is evident that the smallest target segments in the red box of Fig. 16.10c are not visible at all, while these get resolved when observed through the lens (see the area in the red box in Fig. 16.10d) [33].

The integrated optical stretcher (Fig. 16.1) consists of a central microfluidic channel (width 100 µm, depth 90 µm) which is crossed perpendicularly by the light beams coming from the two fibres. After some preliminary experiments, it was clear that the fibres should arrive directly to the microfluidic channel without any plastic wall in front of them. In fact, the power required to trap and stretch cells is sufficient to deform and damage the thin polymeric material that was initially left in front of the fibre to improve the channel sealing. The removable insert was thus modified with the femtosecond laser in order to remove the small wall between the fibre ducts and the central microfluidic channel.

An image of the device is shown in Fig. 16.11. The cross-shaped trace of the laser bonding is clearly visible. It should be noted that the welding trajectory of the fs-laser has been changed several times depending on the specific chip design, further evidencing the simple reconfigurability of femtosecond laser processing. The fibres were glued in their specific holders. It is important to point out that no further alignment of the fibres is needed as they are forced to follow the ducts that are precisely moulded.

Figure 16.12a shows the experimental set-up that was designed to demonstrate the effectiveness of the integrated optical stretcher. A continuous ytterbium fibre laser (YLD-5 k-1070, IPG Fibertech), characterised by an emitting power up to 5 W at 1070 nm, was used as light source. The beam coming from the laser was split into two branches by means of a 50–50% fibre coupler, all the components were single mode at the working wavelength as well as the spliced bare end-fibres (HI-1060-Corning) [27]. The chip was mounted on an inverted microscope, equipped for phase contrast microscopy (Leica), to image the cells in the integrated microfluidic channel. Phase contrast images of optical trapping and stretching of cells were acquired by a CCD camera (DFC310 FX, Leica) through a bandpass optical filter rejecting the infrared light (FGS550, Thorlabs).

We tested the trapping and stretching capabilities of the prototype by using red blood cells. We prepared a solution of 10 µL of blood in 8 mL of hypotonic solution, in which the Red Blood Cells (RBC) acquired a quasi-spherical shape with a radius of ~4 μm. The cell suspension was transported in the microfluidic channel by an external microfluidic pump (MFCS-EZ, Fluigent). The value of the cell speed was set in the 10–50 μm/s range for an easy imaging of the flowing cells. RBCs optical trapping was achieved with an estimated optical power at each fibre output of about 30 mW. When a single cell was stably trapped in the microchannel it could be stretched along the trap axis by simultaneously increasing the optical forces applied to the cell by the two counter-propagating beams [27]. Figure 16.12 shows the image of a trapped RBC inside our chip (b), an increase of the optical power in the two fibres clearly produced the stretching of the cell (c). The current device suffered from the limitation of the surface roughness (measured to be ~600 nm), which did not allow to achieve optimal quality images. This limit could be overcome by making the chloroform vapour treatment to the channel before welding it to the top slab, as explained in Sect. 16.5.3. Figure 16.12d shows the image of a stretched RBC inside a microfluidic channel after the chloroform treatment, indeed the imaging performance was extremely improved. With this image quality, it will be possible to digitally analyse the deformability of the trapped cells.