Functionalized glass coating for PDMS microfluidic devices Adam R. Abate, Daeyeon Lee, Christian Holtze, Amber Krummel, Thao Do, and David A. Weitz* Microfluidic devices can perform multiple laboratory functions on a single, compact, and fully integrated chip. However, fabrication of microfluidic devices is difficult, and current methods, such as glass-etching or soft-lithography in PDMS, are either expensive or yield devices with poor chemical robustness. We introduce a simple method that combines the simple fabrication of PDMS with superior robustness and control of glass. We coat PDMS channels with a functionalized glass layer. The glass coating greatly increases the chemical robustness of the PDMS devices. As a demonstration, we produce emulsions in coated channels using organic solvents. The glass coating also enables surface properties to be spatially controlled. As a demonstration of this control, we spatially pattern the wettability of coated PDMS channels and use the devices to produce double emulsions with fluorocarbon oil. Microfluidic devices consist of networks of micron scale channels that are engineered to perform specific functions. Miniaturization of the channels allows several functions to be integrated onto a single “lab-on-a-chip” microfluidic device(Whitesides 2006). This allows the devices to perform very sophisticated tasks, such as sorting analytes(Ahn et al. 2006; Fidalgo et al. 2008), cells(MacDonald et al. 2003), and worms(Chung et al. 2008), performing combinatorial chemistry(Pregibon et al. 2007), crystallizing proteins(Gerdts et al. 2006), detecting minute concentrations of DNA with ultra sensitive PCR(Cady et al. 2005; Beer et al. 2007), using bubbles for fluidic computing(Prakash and Gershenfeld 2007), as well as a host of other applications(Whitesides 2006). One class of microfluidics that is particularly useful for analysis of chemical and biological systems is droplet microfluidics. With microfluidics, picoliter drops can be formed, merged, and sorted at kilohertz rates(Pipper et al. 2007; Shah et al. 2008; Teh et al. 2008). The drops can serve as individual compartments for chemical reactions(Teh et al. 2008). This combination of speed and containment is very useful for high- throughput screening(Warrick et al. 2007; Guo et al. 2008), useful for the as the discovery of new drugs, the selection of high efficiency chemical catalysts, and the directed evolution of enzymes and cells(Teh et al. 2008). However, microfluidic devices can be quite complex, and their fabrication can be quite difficult. For example, fabrication of glass etched devices requires sophisticated lithographic techniques that are difficult and expensive. Fabrication of milled plastic or metal devices is simple and relatively inexpensive, but the resolution is poor and miniaturization of the fluidic components is difficult(Duffy et al. 1998; Whitesides 2006). By contrast, soft-lithography, PDMS devices can be easily, quickly, and inexpensively fabricated with superb resolution(Duffy et al. 1998; Whitesides 2006). This ability to easily fabricate sophisticated devices has revolutionized the study of microfluidics, particularly in academic labs in which the turn around time must be short. However, PDMS devices also have several significant drawbacks that limit their usefulness for many applications. PDMS is a delicate elastomer that is degraded by common chemicals(Lee et al. 2003; Rolland et al. 2004). Even when cured, PDMS remains permeable to liquids and gases(Lee et al. 2003; Rolland et al. 2004); this limits control and can interfere with reactions in the channels. Small molecules can diffuse into PDMS walls, fouling channel surfaces, and altering device behavior(Roman et al. 2005). Organic solvents, such as toluene and chloroform, are necessary for many applications, including the formation of vesicles and the syntheses of drugs; however, these chemicals swell PDMS, collapsing microfluidic channels, and significantly degrading the device performance(Lee et al. 2003). The poor chemical compatibility of PDMS is, therefore, a major challenge that limits its applicability to lab-on-a-chip microfluidics. This has stimulated the development of new materials that are more chemically robust(Rolland et al. 2004). Alternatively, attempts have been made to increase the robustness of PDMS by modifying its surface properties. For example, by infusing PDMS surfaces with metal-oxide precursors, the diffusion of small molecular weight dyes can be reduced(Roman et al. 2005). By coating PDMS slabs with poly(urethaneacrylate), swelling due to organic solvents can be retarded(Lee et al. 2006). In addition to poor chemical compatibility, PDMS devices are also very difficult to functionalize to control surface chemistry, which is a significant limitation for many applications. For example, in biochemical applications, the microchannels must be coated with proteins to reduce adsorption of reagents to the channel walls. In droplet microfluidics, the interface must be functionalized to control wettability, to ensure that drops of the desired phase can be formed(Anna et al. 2003; Seo et al. 2007). In other applications, the channel properties must be controlled spatially, so that different regions of the device have different properties. For example, the fabrication of sensing devices requires spatial functionalization of the devices with specific chemical groups(Chiu et al. 2000; Rossier et al. 2002; Chen and Lahann 2005). The formation of multiple emulsions requires that the interface have distinctive wettability in different regions of the microfluidic device(Seo et al. 2007). However, spatial functionalization of PDMS is very difficult, and current methods result in channels with only marginal contrasts in wettability and of limited usefulness(Seo et al. 2007). This has stimulated the development of devices fabricated in new materials that can be more easily functionalized. Alternatively, hybrid devices can be fabricated consisting of PDMS channels bonded to a glass plate; the glass plate can be readily functionalized to control surface properties(Li et al. 2007). However, the remaining PDMS faces of the channels remain un-functionalized, so that control is limited, and the devices are still vulnerable to fouling and swelling. Indeed, glass possesses a number of attributes that make it an optimal material for microfluidic devices. Glass is extremely chemically robust: it is resistant to corrosion and fouling, does not swell, and is compatible with a wide variety of chemicals, including organic solvents. Glass can also be functionalized to control surface properties, to graft desirable chemical groups to the surface or to spatially control wettability(Prakash et al. 2007). For example, glass capillary devices can be functionalized to spatially control wettability and can form double and triple emulsions(Utada et al. 2005; Chu et al. 2007), even using organic solvents(Chu et al. 2007). However, glass devices are difficult to fabricate. Glass capillary devices require manual tip pulling to form the drop making nozzles and hand alignment to assemble the devices, tedious processes that are difficult to automate. Glass capillary can only be made to perform a small set of functions, such as forming drops. An optimal system would combine the simple fabrication of PDMS devices with the robustness and control of glass.