Fabrication and bonding of thiol-ene-based microfluidic devices

Materials science and engineering are key elements in the development of lab-on-a-chip applications [1, 2]. Although some cleanroom-based microfabrication techniques allow for low-cost mass production of a high number of identical chips, rapid prototyping approaches, such as soft lithography, are often preferred for benchtop testing of new designs. Thus far, poly(dimethyl siloxane) (PDMS) has been the material of choice for rapid prototyping of microfluidic devices [3]. However, the material properties of the cross-linked PDMS limit its use in advanced bioanalytical applications. For example, PDMS is prone to protein fouling as a result of nonspecific interactions [4], to swelling upon exposure to organic solvents [5], and to absorption of hydrophobic small molecules [6]. Most prominently, owing to its low elastic modulus, the use of PDMS is often limited to relatively low working pressures of about 1 bar [7].

Thiol-ene chemistry [8] and ready-made commercially available thiol-ene formulations have been used increasingly as PDMS substitutes in soft lithography [9–13]. Thiol-ene chemistry is an attractive approach since it allows extremely rapid UV curing through free-radical polymerization even at ambient temperature and pressure [8]. Moreover, the reaction can be initiated without added photoinitiators to yield quantitative polymerization and a uniform cross-linking ratio [14]. The cross-linked thiol-enes are characterized by their strong adhesion to metal and glass substrates [15, 16], good solvent resistance [16–18] and tunable mechanical strength [9]. The surface and bulk properties of thiol-enes are easily tuned toward a number of different applications by simply changing the monomer structure and composition [8, 19, 20]. Even though the radical thiol-ene reaction was introduced as early as the 1920s [21], its recent application in photolithography and microdevice fabrication has attracted a lot of renewed interest. Thiol-enes have been used, for instance, in fabrication of soft lithography stamps [22], micropumps [23], solid–liquid core waveguides [24], functional polymer beads [25] and flow cells for surface plasmon resonance imaging [18]. The fact that the thiol-ene polymerization reaction does not suffer from oxygen inhibition also allows for photolithographic patterning of high-fidelity, high-aspect ratio structures [10]. In addition to the various commercially available thiol and allyl monomers, many research groups have utilized commercial, thiol-ene-based optical adhesives (NOA61 and NOA81, Norland Optical Adhesives, Norland Products, Cranbury, NJ, USA) to prepare a variety of multilayer microfluidic chips by UV-photopolymerization and lamination [16–18, 26–29]. However, the use of NOA adhesives does not allow as much flexibility in the tuning of material properties compared to in-house mixture preparations using pure thiol and allyl monomers. The NOA adhesives have also been reported to suffer from oxygen inhibition, similarly to acrylate-containing resins [30], which results in a thin, uncured residual layer on top of the cross-linked polymer when exposure is performed under ambient conditions or through a gas-permeable PDMS mold. Instead, the thiol-ene reaction is generally insensitive to ambient air/oxygen and moisture, and thus, allows for complete curing [14].

This study focuses on the effects of thiol-ene bulk composition on the thiol-ene to thiol-ene bonding quality. A consequence of tuning the thiol-ene monomeric composition in order to obtain desirable chemical and physical (surface) properties is the impact of these changes on the bonding of microfluidic devices. Even though bonding is one of the most crucial fabrication steps, to our knowledge no previous studies exist on thiol-ene to thiol-ene bonding. The purpose of the study is to examine the different thiol-ene bonding possibilities and systematically characterize the bonding strengths (i.e. pressure tolerance) for different thiol-ene polymer combinations. We also take advantage of the electroosmotic flow supported by the thiol-ene surface to demonstrate rapid capillary electrophoresis (CE) separation on a microchip prepared by bonding two stoichiometric thiol-ene parts.

2.1. Materials and reagents

2.2. Thiol-ene chip fabrication and surface characterization

The thiol-ene chips were made from commercially available monomers by mixing either trithiol with triene or tetrathiol with triene in varying ratios. The different ratios used were calculated with respect to the amount of free thiol and allyl (ene) groups in the monomer structure. No photoinitiator was added to the monomer mixture so that the effects of pure thiol-ene chemistry on surface properties and on bonding quality could be examined. After mixing, the monomer mixture was degassed under vacuum for 5–10 min. The monomer mixture was poured onto the PDMS mold, which incorporated the microchannel pattern, and the mold cavity was sealed with a PDMS lid (figure 1(a)). The PDMS mold was fabricated from Sylgard 184 base elastomer and curing agent mixed in a mass ratio of 10:1, degassed under vacuum for 30 min and cast in a micromilled PMMA master. The PDMS was cured at 80 °C for ∼3 h.

Zoom In Zoom Out Reset image size

The radical thiol-ene reaction was initiated by exposing the monomer mixture through the PDMS cover layer to UV light (Dymax 5000-EC Series UV curing flood lamp, Dymax Corp., Torrington, CT, USA, ∼40 mW cm⁻² at 365 nm) (figure 1(b)). For all thiol-ene compositions, an exposure time of 4 × 1 min with 1 min pause in between exposures was used. In this manner, the temperature rise in the UV chamber leveled off at about +50 °C (ΔT∼30 °C from room temperature). After curing, the cross-linked thiol-ene chips were gently peeled off from the PDMS mold. The functional groups on the cross-linked thiol-ene surfaces were identified based on attenuated total reflectance Fourier transform infrared (ATR-FT-IR) spectroscopy (Spectrum 100 FT-IR, Perkin Elmer, MA, USA).

For the CE chips, a standard separation chip design incorporating an 80 mm long separation channel with double-T (distance between side arms: 100 µm) injection cross was used. The cross-sectional dimensions of the microchannel were 50 µm × 20 µm (w × h). The CE chips were prepared from stoichiometric trithiol-triene using an SU-8 master for PDMS molding and the curing time was 7 min.

2.3. Thiol-ene to thiol-ene bonding

2.4. Pressure measurements

The bonding strengths of the thiol-ene chips were determined by way of measuring the pressure at which leakage from the closed channel system was observed, using a setup built in-house (figure 2). The setup consisted of two 10 mL plastic syringes in a mechanical clamp connected (via a single tube) to a polycarbonate chip holder featuring a single inlet as well as a mount for the pressure sensor (40PC150G, Honeywell, Morristown, NJ, USA). Since the thiol-ene chips had only one inlet hole on the cover part, and no outlet, the entire system was closed. Hence, by slowly using the clamp to compress the air in the syringes the pressure inside the entire setup could be increased up to a value of 6 bar above ambient pressure. At higher pressures the various connectors in the setup would begin to fail, leading to inconclusive results. However, it is assumed that withstanding a pressure of 6 bar should be sufficient for all but the most extreme microfluidic applications. To ensure that the channels were not mechanically blocked or misaligned in the holder, a drop of red food coloring agent was applied to the inlet prior to placing the chip in the holder. Since the chip had no outlet, the dye solution did not spontaneously fill the microchannel, but initially remained at the inlet. As the air pressure increased, the air inside the chip was compressed and the red dye drop traveled toward the end of the microchannel. This allowed us to visualize the air entering the microchannel. During the pressure tests, the air pressure was increased in 1 bar increments after which the system was allowed to stay still for a minimum of 20 s before proceeding to the next pressure level. Here, 20 s was found to be a long enough time to distinguish whether or not the chip failed at the applied pressure, since inflation always began within the first 5–10 s and could be easily observed as deformation and/or delamination of the bonded thiol-ene parts. Those chips that tolerated the highest pressure (6 bar) were kept at this pressure for about 5 min. Compared to the shaft-loaded blister test [31] traditionally used for adhesion testing, the advantage of using microchannels (instead of cavities) filled with pressurized air is that the entire chip area can be examined simultaneously and any leakage can be spotted immediately as the pressure very quickly drops and the escaping red dye stains the leakage point. The output of the pressure sensor was recorded with a LabView (National Instruments, Austin, TX, USA) program written in-house.

Zoom In Zoom Out Reset image size

2.5. Microchip electrophoresis

The thiols and allyls used in this study were simple, commercially available forms of trithiol, tetrathiol and triene (figure 3(a)). No photoinitiator was added to the monomer mixture. In this manner, the thiol-ene polymerization process is known to yield approximately 72% and 80% final conversion of thiols and allyls, respectively [14]. Here, the effect of the monomer composition on the strength of a thiol-ene to thiol-ene bonding was tested by varying the ratio of the thiol and allyl functional groups in the monomer mixtures. In all, four different types of bonding combinations were tested: (i) stoichiometric amount of thiol and allyl groups on both surfaces, (ii) complementary excess of thiols and allyls on opposite surfaces, (iii) identical excess of thiol groups on both surfaces and (iv) identical excess of allyl groups on both surfaces (figure 3(b)). In order to determine how the changes in the monomer ratio were reflected in the surface chemistry of the cross-linked thiol-enes, the UV-cured chips were also characterized by surface sensitive ATR-FT-IR spectroscopy. It was observed that the excess of thiol or allyl functional groups (in the bulk polymer composition) could easily be distinguished via the spectra of the cross-linked thiol-ene surfaces (figure 3(c)). This shows that the surface chemistry is governed by the initial monomer ratio.

Zoom In Zoom Out Reset image size

Figure 4(a) shows that the bonding of stoichiometric substrates (type (i) in figure 3(b)) could withstand pressures up to the system cut-off value at ΔP_max equal to 6 bar. However, as the percentage of excess thiols is increased, the bonding strength starts to weaken. This is clearly the case for both the thiol-to-allyl and thiol-to-thiol bond types (i.e. type (ii) and (iii) in figure 3(b), respectively). In all cases, the bonding failed because of the mechanical deformation of the chip caused by inflation. However, this behavior was not observed for the allyl-to-allyl bonding combination (type (iv) in figure 3(b)). Even when the excess of allyl functional groups was increased to 100% the bonding showed no sign of weakening (data not shown). These results suggest that the bonding strength is dominated by the stiffness of the bulk polymer, whereas covalent bond formation between the free thiol and allyl groups across the bonded interface plays a less important role. The changes observed in the bonding strength correlate best with the changes in substrate rigidity observed for different thiol-ene compositions. For example, Carlborg et al [9] have reported an order of magnitude change in the elastic modulus (E-modulus) from 250 MPa for thiol-enes bearing 90% excess of thiol groups to 1740 MPa for thiol-enes bearing 30% excess of allyl groups. The large amount of unreacted thiol groups present in the substrates containing an excess of thiol groups results in a soft, flexible material, which is easily deformed under applied pressure. In contrast, when the amount of allyl functional groups exceeds stoichiometric ratios, the cross-linked thiol-ene becomes more and more rigid and thus, the pressure tolerance exceeds the maximum value (ΔP ≥ 6 bar).

Zoom In Zoom Out Reset image size

It has been suggested previously that two thiol-ene layers with complementary excess of thiols and allyls can be covalently bonded based on UV-initiated thiol-ene free-radical polymerization reaction between the thiol and the allyl groups at each interface [9]. The results presented here suggest that the bonding strength of a thiol-ene chip is dominated by the stiffness of the chip layers. This effect was further examined by evaluating a chip where an elastic thiol-ene layer containing 30% excess thiol was sandwiched between two stiff thiol-ene layers containing 30% excess allyl (i.e. a three-layer allyl-thiol-allyl chip). Thanks to the rigidity of the surrounding excess allyl layers, this three-layer combination was able to withstand more than 6 bar. In order to further investigate if the polymer bulk properties (stiffness, cross-linking ratio) govern the bonding quality over the surface chemistry, a new line of chips was fabricated using the thiol-to-thiol bond combination (type (iii) in figure 3(b)). These chips were made using the tetrathiol monomer (figure 3(a)), which bears four functional thiol groups and thus produces a more densely cross-linked structure and stiffer thiol-ene substrates than those made using the trithiol monomer. The bonding strengths of the new stiffer tetrathiol-based chips were compared to those of the trithiol-based chips using the same excess of thiol functional groups on both sides of the bond (figure 4(b)). The results in figure 4(b) clearly show that for chips made of trithiol the bonding strength begins to decrease already at a thiol excess greater than 5%, whereas with stiffer chips made of tetrathiol, up to 30% excess of thiols still provides the maximum bonding strength.

Finally, the effects of storage time and UV exposure on the bonding strength were examined with chips prepared from trithiol and triene, and incorporating 20% excess of either thiol or allyl functional groups. Three different bonding combinations were tested: thiol-to-allyl, thiol-to-thiol and allyl-to-allyl (types (ii), (iii) and (iv) in figure 3(b)). UV exposure after lamination of the cross-linked thiol-ene parts was critical to the bonding quality as it initiates the thiol-ene reaction between free functional groups on both substrates. If UV exposure was not performed, a maximum bonding strength of only 3 bar was obtained for the allyl-to-allyl chips, while the other chip combinations failed at even lower pressure differences (≤1 bar) (figure 4(c)), demonstrating that simple lamination is not sufficient for bonding. The storage time did not have an equally dramatic effect on the bonding strength; still, the best bonding quality was obtained for chips that were bonded immediately after curing (figure 4(c)). The effect of aging (thiol-ene layers stored 24 h protected from light before bonding) was more prominent in the case of thiol-to-allyl and thiol-to-thiol chips (types (ii) and (iii) in figure 3(b)), while being practically negligible in the case of allyl-to-allyl chips (type (iv) in figure 3(b)). The latter combination provided the maximum bonding strength even after the parts were stored for 24 h before bonding. These results suggest that the UV-initiated radical thiol-ene reaction may indeed occur across the bonded interface since UV exposure clearly improves the bonding quality. However, the material stiffness seems to dominate the effect of the cross-linking reaction as the most rigid chips (i.e. allyl-to-allyl chips) were not affected as much as the more elastic thiol-to-allyl and thiol-to-thiol chips.

In all, the results presented here show that most thiol-ene to thiol-ene bonding combinations easily withstand pressure differences of up to 6 bar. The only limitations regarding the bonding strength appear with thiol-enes containing a large excess of thiols because of their relatively high elasticity. In such cases, one can increase the polymer stiffness by choosing a thiol monomer, which has a rigid molecular backbone and which yields symmetric polymer structure with a higher degree of crosslinking (such as tetrathiol). Most importantly, our results evidence that thiol-ene chips can be prepared by bonding not only reverse off-stoichiometric thiol-ene parts (i.e. excess thiols to excess allyls), but also one stoichiometric thiol-ene part to another stoichiometric part as well as off-stoichiometric parts with the same excess of thiols or allyls on both surfaces (i.e. thiol-to-thiol or allyl-to-allyl). Thus the flexibility in bonding various surfaces together allows for accurate control of the microchip properties toward the desired application area. Bonding of two thiol-ene parts with different surface chemistry allows for functionalization of a microchannel wall independent from the rest. This can be extremely useful in biochip applications where site-specific surface immobilization of biomolecules is required, such as in evanescent wave-induced fluorescence spectroscopy [32]. On the other hand, bonding of two similar surfaces is preferred in order to obtain uniform surface properties in the microchannel. For instance, if electroosmosis (or some other surface driven technique) is to be used for fluid actuation, it is advantageous to have the same chemistry on all four microchannel surfaces. Also, most separation systems require uniform interactions (or lack of them) on all surfaces in order to avoid peak broadening. Here, the good separation efficiency obtained on an electrophoresis microchip prepared by bonding two stoichiometric thiol-ene parts was demonstrated by separation of fluorescent model compounds (figure 5). The number of theoretical plates obtained for the model compounds rhodamine 123 and fluorescein was 1635 and 2665 per 5 cm, respectively. The migration times of the model compounds were also repeatable within 4.2–6.0% RSD (n = 6) which evidenced that, thanks to the uniform surface chemistry, the electroosmotic flow was stable from run to run and no leakage occurred.

Zoom In Zoom Out Reset image size

In this work, we studied the effects of monomeric thiol/allyl composition on the bonding strength and surface chemistry of microchips fabricated by thiol-ene free-radical polymerization. Although thiol-ene chemistry is increasingly applied to fabrication of microfluidic and micromechanical components, to our knowledge, this is the first time that the different approaches for bonding were characterized in detail. By using infrared spectroscopy, it was confirmed that the changes in the monomeric thiol/allyl composition altered not only the bulk polymer properties but also the amount of free thiols/allyls on the substrate surfaces. Our results show that thiol-ene chips can be prepared in various combinations, either by bonding two thiol-ene parts exhibiting reverse off-stoichiometry (i.e. thiol-to-allyl) or by bonding two thiol-ene substrates with the same surface chemistry (i.e. stoichiometric-to-stoichiometric, thiol-to-thiol, or allyl-to-allyl). Most bonding combinations tested easily withstood pressures greater than 6 bar. In general, the bonding strength was proportional to the stiffness of the cross-linked polymer so that only the most elastic thiol-ene chips (exhibiting significant excesses of thiols) failed at pressures lower than 6 bar. However, the stiffness, and thus the bonding strength, of thiol-enes prepared with excess thiols was easily increased by using monomers that provided more symmetric polymer structure with a higher degree of cross-linking (e.g. tetrathiol versus trithiol). Such flexibility in bonding is likely to allow for accurate control of the microchannel surface chemistry and thus versatile use of the thiol-ene chips in various different applications. For instance, bonding of two similar thiol-ene parts with uniform surface chemistry on all microchannel walls is advantageous for separation chips. Here, CE separation of fluorescent model compounds was demonstrated on a microchip made of stoichiometric thiol-ene. On the other hand, bonding of two thiol-ene parts with different surface chemistry allows for functionalization of a microchannel wall independent from the rest. This feature would have applications for instance in evanescent wave-induced fluorescence spectroscopy, where only the wave-guiding surface should be functionalized.

This work is part of the NaDiNe EU project (contract 246513) and was supported by the European Commission through the Seventh Framework Programme and by the Academy of Finland (grant 251629).