A syringe pump was required to deliver liquids into a PDMS microfluidic channel since its surface was hydrophobic. However, PEG microchannel did not require such pumping since its surface was hydrophilic and liquids could be delivered by capillary action (Fig. 1). The rigidity and surface roughness of cross-linked PEG substrate made bonding to a cover slide difficult, leading to liquid leaking. As addressed in Methods section, we used an adhesive tape as a cover slide, and were able to resolve both cover bonding and liquid leaking issues without any complicated bonding procedure.

Fig. 2 shows representative, light and fluorescent microscopic images (out of three different sets of experiments) for latex immunoagglutination assays (a and c; taken 3-mm away from the Y-junction) and for injection of fluorescein-labelled BSA (b and d; taken at the Y-junction) after one, four, and ten uses of the same microfluidic channel. Immunoagglutination made the particles to form mostly triplets or larger clumps at high target concentrations (a few tens of μg ml^-1 and beyond), while mostly doublets at low target concentrations 17. Therefore, we defined the microfluidic channel not reusable when those triplets and/or larger clumps could be observed in microscopic images after rinsing. We also defined the microfluidic channel not reusable when three or more bright fluorescent spots (of BSA) could be observed in microscopic images. The PEG microfluidic channels (a and b) were free of such particle agglutinates or BSA after four uses. Noticeable contaminations were observed only after ten uses. However, significant contaminations were observed for the "fresh" PDMS device after four uses. "Aged" PDMS device was a lot worse, showing significant fouling after the very first use (data not shown). This "aged" PDMS was virtually "plain" PDMS, since the hydrophobic recovery gained by oxygen plasma treatment fades away within 48 hours, known as hydrophobic recovery 18. The actual number of reusable time may be much higher for PEG microchannel, since our target concentration (1 mg ml^-1) was relatively higher than those of typical biological assays (in μg ml^-1 to ng ml^-1 scales).

The water contact angles on PEG surfaces were obtained against the number of rinsing performed (Fig. 3). PEG surfaces initially showed very low contact angles (23°). This angle gradually increased and levelled off at 50° after three rinses and beyond. Meanwhile, the contact angles on "fresh" PDMS started at 7° (no rinsing), rapidly increased to 68° at four rinses, and levelled off at 85° after six rinses and beyond. The contact angle of ~60° seems to be a threshold in determining reusability of a microfluidic device for particle immunoassay or passive protein adsorption. As the contact angle on PEG surfaces stayed at 50° (i.e. below ~60°) even after 10 rinses (data not shown), the contaminations on a PEG microchannels after 10 rinses may be attributed to permanent protein fouling on surface cracks and/or dust particles, not by surface hydrophobicity.

This threshold angle of 60° can be correlated to those on protein films. Water contact angles on salt-free, dried protein films were measured as: 75 ± 1° for bovine serum albumin, 67 ± 1° for bovine hemoglobin, and 47 ± 1° for hen egg white lysozyme. These contact angles are comparable to our threshold angle of 60°. Once the water contact angle of microchannel surface exceeds those of proteins, hydrophobic-interactions-induced protein adsorption will be preferred on microchannel surface that will permanently foul the surface 1920.

In order to perform particle immunoassays, antibodies were conjugated to microparticles by physical adsorption as described previously 17. Briefly, 1 ml of 0.02% w/v, 0.92 μm highly carboxylated polystyrene particles (parking area = 10.3 Ų per carboxyl surface group; Bangs Laboratories, Fishers, Indiana, USA) were mixed with 1 ml of 1.023 μg ml^-1, anti-mouse immunoglobulin G (anti-mIgG; catalog number M8642, Sigma-Aldrich Co, St. Louis, Missouri, USA) solution, followed by centrifuging and resuspension (the whole cycle was repeated twice) to eliminate the free antibodies. The surface coverage of the antibodies on particle surface is approximately 33%, which is appropriate in maximizing particle immunoagglutination 21. Target protein was mouse immunoglobulin G (mIgG; catalog number I5381, Sigma-Aldrich). For a comparison purpose, 1 mg ml^-1 fluorescein-labelled bovine serum albumin solution (with fluorescein isothiocyanate; FITC-BSA; catalog number A9771, Sigma-Aldrich) was used to monitor the protein-fouling behaviour within the PEG microfluidic channel. All dilutions were made with 10 mM phosphate buffered saline (PBS; pH 7.4; Sigma-Aldrich).

Fig. 4 (left) shows the layout of a Y-shape PEG microfluidic channel. PEG microfluidic channels were fabricated by drop-dispensing polyethylene glycol diacrylate (PEG-DA) with 1% w/v UV initiator (2,2-dimethoxy-2-phenylacetophenone) on a glass slide. The glass slide was pre-modified with adhesion promoter (acrylic acid dissolved in propylene glycol monomethyl ether acetate; 10 vol.%) 16. This glass slide merely serves as a support for PEG-DA. A stamp made by the popular PDMS molding technique 22 was used to transfer a pattern onto a PEG-DA substrate as shown in Fig. 4 (right). As the PDMS mold did not make a contact with the glass support, microchannels made entirely out of PEG-DA could be fabricated. Microchannels were 100 μm wide and 100 μm deep. The PEG-DA layer on a glass support was 200 μm thick. A removable adhesive tape (Scotch^®, 3 M, St. Paul, Minnesota, USA) was used as a cover for PEG microfluidic channel as shown in Fig. 4 (left). This tape cover could be discarded after each assay.

Particle suspensions and/or protein solutions were introduced to a PEG microfluidic device via capillary action from the liquid droplets (3 μl each) sitting on the inlets, as shown in Fig. 1. Five minutes after introducing the solutions, an adhesive tape was removed and the microfluidic channel was rinsed with deionized water, followed by observation of the inner surfaces of a microfluidic channel with an inverted, light or fluorescent microscope (Nikon Instruments, Tokyo, Japan) (Fig. 1). This procedure was repeated until visually identifiable particle agglutinates (i.e. triplets or larger clumps) or BSA (i.e. three or more bright fluorescent spots) could not be removed from the microfluidic channel by rinsing. Additionally, we used PDMS microfluidic channels (i.e. the most popular device) as negative controls, with the same layout and dimensions of microchannels. Both "fresh" oxygen-plasma-treated PDMS (water contact angle < 10°) and "aged" PDMS (two days of incubation at room temperature after oxygen plasma treatment; water contact angle ~90°) were tested. The fabrication procedure of PDMS microfluidic device can be found in previous publications 1723.

A contact angle/surface tension analyzer (FTÅ200, First Ten Ångstroms, Portmouth, Virginia, USA) was used to measure the contact angles on the PEG surfaces with 5 μl sessile drops of deionized water. To simulate water rinsing of a microfluidic channel, the PEG surfaces were rinsed with deionized water after each contact angle measurement. Sessile drops were placed on the surfaces for 2 min, the same as the liquid exposure time of PEG (and PDMS) microfluidic channels. A single data point was averaged from three different measurements of contact angle on PEG (and PDMS) surfaces. Substrates were thoroughly dried with nitrogen gas prior to contact angle measurement.

Water contact angles on salt-free, dried protein films were also measured using the same instrument. Bovine serum albumin (catalog number P-7656, essentially salt-free, Sigma, St. Louis, Missouri, USA), bovine hemoglobin (catalog number H-9891, essentially salt-free, dimethylated and primarily methemoglobin, Sigma), and hen egg white lysozyme (salt-free, catalog number 10 837 059 001, Roche, Germany) were used as model proteins. 15 μl of 1 mg ml^-1 albumin, haemoglobin or lysozyme solution was deposited on a precleaned glass microscope slide (catalog number 12–552, Fisher Scientific, Pittsburgh, Philadelphia, USA), and stored in a nitrogen-purged desiccator for more than a week. 5 μl droplets of deionized water (from Millipore's Simplicity, resistivity > 18 MΩ cm) were automatically dispensed by the same contact angle analyzer, and deposited on the protein films. Contact angles were measured right after a droplet stops vibrating and forms a perfect spherical shape. This was normally achieved within 33 ms (images were captured every 33 ms). Sometimes the contact angle kept decreasing significantly over time, probably due to the absorption of solvents into the films. We simply eliminated such data from our experimental set.