Excimer laser micropatterning of freestanding thermo-responsive hydrogel layers for cells-on-chip applications

In order to produce PNIPAAm thermo-responsive hydrogels, we adopted a synthetic process based on free radical polymerization [20]. All chemicals were purchased from SIGMA Aldrich, unless stated otherwise. The principal drawback in many applications based on the use of PNIPAAm consists in the typical poor mechanical properties of the polymer making it difficult to handle [21]. To overcome this limitation, reaction time and temperature, monomer concentration, solvent type and initiator concentration were varied in order to obtain a transparent hydrogel with suitable mechanical features. The best trade-off was found for a polymerizing mixture obtained by adding reagents with relative amounts as follows: N-isopropylacrylamide (NIPAAm, 377.4 mg, 3.34 mmol), ethanol (EtOH, 558 µl), MilliQ or de-ionized water (H₂O, 110.2 µl), ethylene glycol dimethacrylate (EGDMA, 34.9 mg, 0.176 mmol), tetramethylethylenediamine (TEMED, 17.1 mg, 0.147 mmol) and 10% w/w ammonium persulfate aqueous solution (APS, 24.1 µl, 0.0103 mmol). Different concentration of APS as a redox initiator was tested in the polymerization reaction, and it was observed that the minimum amount required to induce cross-linking has to be used in order to avoid hydrogel inhomogeneity. Reaction time and temperature were varied in the range 3–48 h and 4–28 °C, respectively, and no evident correlation with the final hydrogel film's properties was found. The NIPAAm:EGDMA ratio is 1:0.01 mol/mol, while the monomer:initiator ratio is set as 0.6%. Once the reaction takes place, it takes 4 h for gelation and almost 15 h for completing the polymerization and cross-linking of the hydrogel. A scheme of the reaction leading to the molecular unit forming the polymeric network is depicted in figure 1.

Zoom In Zoom Out Reset image size

The swelling ratio of PNIPAAm, defined as the weight of the swollen hydrogel in respect of its dry weight [22], was measured with the blot and weigh method [23] using a precision balance to weigh moulded cylindrical samples (15 mm diameter, 1 mm height). The swelling ratio resulted to be

$$\begin{equation} Ws/Wd = 1.5 {,} \end{equation} \tag{ 1 }$$

where Ws is the weight of the hydrogel in the swollen state, while Wd is its weight in the dry state. Although similar moulding protocols were mentioned by other authors [24], the compression-based fabrication method we developed for producing freestanding PNIPAAm thin films has not been described previously. The technique was designed to achieve an appropriate flatness all over the moulded layer length, such that the laser machining could be carried out identically in every region of the film without causing blurs or distortions of the holes' profile due to local thickness variations. Furthermore, the cross-linked polymer had to be physically anchored to the substrate strongly enough to avoid spontaneous detachment and allowing at the same time a relatively ease in being removed from its support after the machining. Repeatability, thickness accuracy and the use of a rather minimal setup in terms of materials and equipment were also required in the perspective of developing a process with scale-up features. The three components of the moulding system employed are as follows: (i) flat 2 mm thick polyvinyl chloride (PVC) rectangular substrates (15 × 5 cm²), (ii) hollow silicone moulds with various thicknesses (75–300 µm) and (iii) 5 mm thick polished aluminium rectangular lids (the same surface area as of the plastic components). PVC layers were treated on one surface with a wet roughening process, by continuously pouring MilliQ water over the material while using sandpaper at the same time in order to promote finest levels of surface roughness (200 and 400 grits were used). Silicone moulds were produced in-house using Daw Corning Sylgard 184 polydimethylsiloxane (PDMS). The protocol for flat PDMS thin sheet realization consists in pouring the degassed viscous polymer (elastomer:curing agent = 10:1 w/w) over a flat aluminium slab, in a region delimited by hollow brass-based foils, which are properly positioned over the metallic substrate; the system is then closed with another aluminium layer and compressed using 30 kg cylindrical brass weights. The curing of the silicone takes place in 30 min at 120 °C over a preheated hot plate. This process allows a precise control over the PDMS films' thicknesses, which results within 10% of the brass-based spacers' one. Once the elastomeric sheets are removed, they are punched or shaped with a central hollow region with a cutter. The technique for the production of PDMS spacers is shown in figure 2.

Zoom In Zoom Out Reset image size

Once the equipment is set up, PDMS moulds are placed over the rough PVC substrate in such a way that adhesion between the two materials is guaranteed. The hydrogel polymerizing mixture previously described is then injected with a calibrated pipette in the hollow region defined by the mould. In order to realize PNIPAAm films with thicknesses ranging from 100 to 300 µm and with a rectangular surface area equal to 10 × 2 cm², 0.5–0.7 ml of solution is sufficient, taking in account leakage phenomena that inevitably occur at the PVC/PDMS interface. Before the injected liquid fully spreads over the PVC, the polished aluminium lid is pressed over the moulding system and compression is then promoted by positioning a cylindrical brass weight over the top layer. This procedure ensures that, for the whole time required for the hydrogel to fully solidify, the system is properly sealed. When disassembly of the moulding apparatus takes place, the different roughness of the substrate and lid allows the film to physically stick to the PVC, while the aluminium component can easily be removed. When employing weights between 2 and 5 kg, the resulting hydrogel films exhibit a thickness equal to the PDMS mould's one within 10–20%. The moulding scheme for PNIPAAm-based thin film realization is reported in figure 3.

Zoom In Zoom Out Reset image size

To further reduce the error on dry PNIPAAm films' final thicknesses, a similar moulding apparatus which relies on a screw-based compression system has been realized; with this improvement, accuracy in thicknesses could reach values below −10%. The particular choice of the materials employed in the moulding system is the result of a systematic experimental analysis carried out on a wide range of possible candidates, both for the substrate and lid, including glass, polystyrene, polymethylmethacrylate (PMMA), rough aluminium and others (data not shown). The developed manufacturing protocol presents features that have potential to be scaled up for mass production.

The through-hole machining on the dry PNIPAAm films by means of the KrF excimer laser (computer-controlled Exitech series 7000, pulsed beam, UV, 248 nm) has been carried out in air, at room temperature. During fabrication, dry hydrogels are still attached to the PVC substrate, in order to ensure a reasonable flatness of the work-piece. The focal distance of the laser is fixed by the system's optics and the sample position is electronically controlled by moving the work-piece stage in the direction perpendicular to the plane defined by the stage itself, such that the projected pulsed beam results focused on the hydrogel top surface.

The focal length and diameter of the demagnification lens are equal to 49.5 and 10 mm, respectively, while its numerical aperture (NA) is 0.2. Working with these optical parameter values allows not only to achieve a high resolution of the machined features (R∼ λ/NA = 3.1 µm), but also to neglect the relatively low differences in the thickness values of the hydrogel film over its surface area (<10%), which do not affect the machining process in terms of reproducibility of the pattern present on the masks due to the depth of field (DOF) of the projection system (DOF ∼ 40 µm). Focal position for each sample was determined by firing the laser on the hydrogel with ten shots per mark over eight different regions located in the film, moving the sample stage elevator 0.1 mm in each step. This procedure sometimes required refinement around a specific position determined to improve the quality of the pattern produced; in this process, the stage elevator is moved 50 or 20 µm off each mark. The projection mask is a 250 µm thick aluminium (4 × 4 cm²) square foil and the pattern design is a 9 × 9 array of equally spaced through-holes with 200 µm diameter and was produced by chemical etching. Two different masks were also realized with the same technique, with hole diameters being 100 and 300 µm, respectively. Thus, since the demagnification of the laser system is 1:10, the entrance hole diameters on the hydrogel are expected to be 10, 20 or 30 µm, respectively, depending on the mask employed. A schematization of the machining apparatus is reported in figure 4. The number of holes and the value of their diameters have been chosen in order to achieve a configuration of 81 tapered micro-capillaries, separated by a centre-to-centre distance of 100 µm, having the cross-sectional dimension comparable with that of a sphere of 10, 20 or 30 µm.

Zoom In Zoom Out Reset image size

Proper values of the sample-focused beam mutual distance as well as the fluence threshold beyond which PNIPAAm etching is actually achieved have been experimentally evaluated by analysing the samples with direct microscope imaging. The best choice for achieving good resolved features resulted in laser output energy per pulse equal to 250 mJ, with repetition rate fixed at 5 Hz and number of shots ranging from 400 to 3200. Focal position varied from sample to sample according to hydrogel thickness. An example of a machined PNIPAAm dry film (200 µm thick, mask holes' diameter 200 µm) is reported in figure 5. The set of chosen parameters is a trade-off between the overcome of the ablation threshold and the minimization of debris traces, which are present in a relatively small amount (figures 5(c) and (d)) and which did not show to depend on the number of shots delivered, until this value is set below 2400. Moreover, the micrometre scale spatial resolution and process repeatability achieved with this fabrication protocol, as well as the relatively low cost of the materials involved, render this hydrogel thin films' patterning suitable for further scale-up manufacturing system developments.

Zoom In Zoom Out Reset image size

In order to produce the freestanding micro-patterned PNIPAAm films, a chemical treatment has been carried out on the dry samples. Experimental analysis evaluating different degrees of hydrogels' swelling promoted by a number of organic solvents (methanol, ethanol, acetone, 2-propanol) and aqueous solution containing the same (solvent:water <30:70 v/v) has led us to determine which liquid environment constitutes the best option for detaching the films from PVC without any damage. By soaking the hydrogels anchored to PVC in ethanol, PNIPAAm films were completely detached from the substrate without any supplementary mechanical operation, as a result of the polymer's swelling. Both the in-plane and normal stresses generated by the elastic relaxation of the polymer chains due to the slow incorporation of the solvent in the matrix contribute to overcome the weak work of adhesion at the layer/substrate interface and allows a homogeneous detachment along the whole surface area of the film in a 1 h time window. Furthermore, the liquid penetrating at the forming interface acts as a lubricant to facilitate the process avoiding fractures or failures. To wash the unreacted compounds, films were then rinsed in an ethanol aqueous solution (ethanol:water = 70:30); complete swelling in water is reached by decreasing ethanol concentration by the 20% each hour. Films were then stored at room temperature overnight previous to metrology and thermo-response testing.

Holes' tapering in the swollen state at room temperature was first measured by means of confocal microscopy. The microscope employed is an inverted Leica DMIRE2 HC Fluo TCS 1-B-UV, equipped with Ar and HeNe lasers. The main idea underlying these metrology experiments is to incorporate a small fluorescent dye into the hydrogel matrix, by soaking PNIPAAm samples in a proper aqueous solution containing the molecule of interest and then collect the fluorescent signal from the samples. Within this frame, only the dye-free through-holes would not contribute to the overall fluorescence emission, thus appearing as dark circles in the in-plane (fixed z) images acquired. By processing the set of stacks using ImageJ software, a 3D reconstruction of the holes is then available. More specifically, samples were soaked in an aqueous solution of rhodamine B (20% v/v) overnight, and then briefly rinsed in pentane before analysis in order to remove residual water containing fluorescent dye traces from within the holes. During imaging, PNIPAAm specimens were positioned between two glass slides; in order to avoid dehydration, the acquisition must take place in a time within 30 min after film's positioning on stage. Moreover, the entrance and exit holes' diameters are also measured by means of the optical inverted microscope imaging in order to achieve supplementary information on the holes' tapering; in these set of measurements, PNIPAAm thin films are kept in a water bath at room temperature during observation.

To investigate holes' shrinking due to PNIPAAm freestanding films' dehydration when temperature is raised over 32 °C, an experimental apparatus has been employed as shown in figure 6.

Zoom In Zoom Out Reset image size

Hydrogel patterned films are immersed in a heated water bath (2.5 ml), whose temperature can be precisely regulated employing a system made up of a proportional integral derivative (PID) controller (CAL9500), a heating element and a k-type thermocouple (KTC). Samples are analysed by optical microscopy in real time as temperature is changed over a range of more than 20°, from 18 to 39 °C. To hold the specimens in place during imaging, a 150 mm thick glass slide is placed over the hydrogels, in order to guarantee a mild compression so that the films would not endure mechanical constraint. A 120 µm thick transparent heater (MINCO, model H6700R9.0, 9 Ω, power supply 5 V) has been embedded in a moulded PDMS box and connected to the output port of the PID controller. The heating element is positioned at the bottom of the box, 1 mm underneath the container surface, in order to ensure homogeneous heating of the water bath. The thermocouple is held in place suspended in the liquid's bulk and the measured temperature values are continuously processed by the PID system. A feedback loop between the heater and KTC ensures accurate control over the water's temperature (±0.5 °C), which can be directly set from the controller. Each sample is analysed by carrying out multiple temperature cycles to observe the reversibility of the process.

After swollen in water, freestanding films were punched in circular discs (6 mm diameter), presenting the holes' array positioned at their centre. According to the expected swelling ratio, the machined through-holes showed an increase in the diameter by 50% with the dry state (figure 7).

Zoom In Zoom Out Reset image size

After the swelling procedure was carried out, debris traces around the machined holes were partially washed away, while the edges of the structures remained approximately as they were in the dry state, reflecting the actual features of the machining process (figure 7(b)).

Results on the holes' tapering reconstruction by means of confocal microscopy on a water swollen 100 µm thick film, machined with the 200 µm diameter hole mask, are shown in figure 8.

Zoom In Zoom Out Reset image size

The acquisition height step (distance between individual in-plane images) was set as 0.5 µm. The reported picture in figure 8 shows a negative image of the processed stacks, where the through-holes appear red labelled over a dark background. A massive difference between the entrance and exit hole diameters is clearly observed. A measure of the two diameters, averaged over the holes in the array, gives the values of 30 and 10 µm for entrance and exit holes, respectively. The tapering angle could then be calculated to be 84°. To further investigate the holes' shape, optical inverted microscope images have been recorded both for entrance and exit holes. With this simple technique, a reliable 3D reconstruction of the tapering could be carried out, as reported in figure 9 for a 250 µm thick water-swollen freestanding PNIPAAM film, machined using the 300 µm diameter holes' mask (the number of shots equal to 3200).

Zoom In Zoom Out Reset image size

Figure 9 shows a PNIPAAm 100 µm thick film undergoing a volumetric phase transition when the temperature is set to 18, 25, 33, 37 and 39 °C. Machining has been carried out on the sample using the 200 µm diameter holes' mask.

The polymer heating rate has been driven by the temperature ramp set by the PID system, which is approximately 1 °C min^–1. No appreciable changes in the sample dimensions or in their transparency have been observed for temperatures below 27 °C. A slight decrease in the disc's dimensions (∼10–15% of its initial diameter) and a mild opacity of the hydrogel started along the imposed heating rate between 28 and 30° C, followed by an abrupt shrinkage that brought the polymer to its equilibrium collapsed state in few seconds after 32 °C is reached. It has been observed that when the hydrogel's temperature is driven beyond the transition point, the pattern's shrinkage occurs along the whole film's deswelling process, thus leading to a significant reduction in through-hole size. The average entrance holes' diameter has been measured to decrease from 30 to 15 µm for the sample imaged in figure 10, and compatible features could be observed for the exit holes, according to the tapering values. Similar behaviour has also been observed for samples with different thicknesses (250, 150 and 100 µm) which were machined employing different projection masks; for the thickness range explored in these tests, no effects on the shrinking behaviour due to size differences have been observed and the shrinking degree of the micro-structures resulted invariant. Supplementary thermo-sensitivity experiments have been lately conducted on thinner micropatterned PNIPAAm layers, approaching a minimum thickness of 50 µm; the deswelling mechanism has still shown the same phenomenology, preserving holes' diameter changes. Furthermore, the process has shown reversibility when multiple cycles have been carried out.

Zoom In Zoom Out Reset image size

The results about the swelling–deswelling kinetics over each cycle showed identical characteristics in terms of response rate and relative hole dimension changes when the temperature was varied from 18 to 39 °C, but exhibited a slower reswelling process for the film to recover its initial state; this could be interpreted as a consequence of the reduced mobility of the hydrophobic polymer chains in their globular conformation in respect of the coil-like hydrophilic state, as pointed out in [25]. We believe that the freestanding character of the fabricated films is not only responsible for the homogeneous and substantial isotropic restriction of the machined geometrical features, but also promotes buckling-free deformation of the through-holes and avoids significant distortion phenomena affecting their shape, at least at the mesoscopic scale. These experimental observations can be compared with what has been reported in the literature concerning the behaviour of mechanical constrained thermo-responsive micro-printed PNIPAAm-based films. Khademhosseini et al [26], as well as Eiichi et al [27], realized micro-patterned PNIPAAm grafted thin films, chemically anchored to solid substrates; due to the in situ fabrication methods employed, the deswelling mechanism is confined only to the thickness' direction for these systems and an increase in the holes' diameter when the temperature is raised beyond 32 °C is observed. Conversely, fully three-dimensional swelling/deswelling, lacking radial stresses at the film–substrate interface, allows the temperature-triggered smart freestanding surfaces to completely transfer their volumetric transition behaviour to the micro-features designed by the excimer laser. This means that a reduction in the whole film's size does not promote an increase in the holes' geometrical dimension but rather favours their shrinkage, according to the overall macroscopic deswelling process. These features render freestanding micro-patterned PNIPAAm films promising and interesting tools to be employed in the frame of properly designed lab-on-chip systems which rely on the use of any size-dependent sorting operation, such as filtering, or which are engineered for promoting cell sorting and cell isolation mechanisms for biotechnological applications. In the latter perspective, we carried out preliminary tests in order to explore the possibility of integrating these smart surfaces with other suitable components based on thermoplastic materials.

A three-layer microfluidic device has been designed to determine whether PNIPAAm thin films could endure compression-based bonding when implemented as an interface between two thermoplastic material layers.

The principal focus of these preliminary tests is to identify a range of pressure, determined by the multilayer assembly, which can be endured by a water-swollen PNIPAAm film operating as a gasket element when interposed between two harder polymeric layers. In this regime, no liquid leakages at the different materials' interface should occur when fluid is perfused at relatively low-pressure values (from few Pa to fractions of KPa) on top of the hydrogel surface. Flow rate values were chosen as the independent variable to impose the fluidic regime, being reasonably representative for a wide class of microfluidic perfusion devices [28]. The results of this set of experiments reflect a purely mechanical characteristic of the material synthesized under the condition previously described, rather than aiming at defining a proper operating behaviour which would be strictly depending on its thermo-responsitivity or micro-structured features.

The average duration of each test is around 30 min and no further characterization has been carried out to explore long-term stability effects on the hydrogel components so far; this is also due to the fact that the main idea behind the prototype of cell-sorting chip based on the smart films developed is conceived to be disposable and operating in the time scales of few tenths of minutes. The test micro-system is represented in figure 11(a), while in figure 11(b), a photograph of the device is reported.

Zoom In Zoom Out Reset image size

Polymethylmethacrylate (PMMA) 3 mm thick slabs were machined using a numerical control micromilling process in order to realize microchannels, fluidic ports, as well as through-holes for hosting screws; a non- machined PNIPAAm thin film with a surface area of 2.5 × 3.5 cm² could be positioned over the bottom flat PMMA component, in a region defined by a hollow 75 µm thick metallic spacer. The top PMMA layer is then closed over the film; thus, the micromilled channels (0.5 × 1 mm² rectangular cross section) present over its bottom surface could get sealed on the hydrogel platform. Two aluminium jigs were also realized and employed to properly compress the three-layer system's components by controlling the screwing (figure 11(b)); in this configuration, 100 µm thick films were thus compressed by 25% of their initial thicknesses.

The sealing pressure of the device is fixed by the screw configuration and is equal to 0.25 MPa. Liquid leakage tests at the interface between the PNIPAAm film and PMMA top layer were then carried out by perfusing the microchannel with a concentrated red dye aqueous solution (E122-Azorubine, 30 mmol) continuously delivered via a diaphragm micropump, at a rate varying from 0.1 to 5 ml min⁻¹; the volumetric flow rate values (Q) employed as the independent input variable and the corresponding pressure differences ΔP expected to be developed in the conducts are reported in table 1. ΔP values are calculated according to the fluidic resistance of each microchannel [29].

Within this volumetric flow rate range, the generated pressure values were suitable to avoid liquid leakages (figure 11(b)) as observed by real-time optical microscope imaging. Moreover, after each perfusion test, the interface regions between the hydrogel and PMMA as well as the sink channels surrounding the perfused microstructures were observed by increasing the objective magnification in order to validate the previous analysis.