Scalable attoliter monodisperse droplet formation using multiphase nano-microfluidics
- First Online:
- Cite this article as:
- Shui, L., van den Berg, A. & Eijkel, J.C.T. Microfluid Nanofluid (2011) 11: 87. doi:10.1007/s10404-011-0776-7
- 1k Views
We demonstrate a robust method to produce monodisperse femtoliter to attoliter droplets by using a nano-microfluidic device. Two immiscible liquids are forced through a nanochannel where a steady nanoscopic liquid filament forms, thinning close to the nanochannel exit to a microchannel due to the capillary focusing. When the nanoscopic filament enters the microchannel, monodisperse droplets are formed by capillary instability. In a certain range of physical parameters and geometrical configurations, the droplet size is only determined by the nanochannel height and independent of liquid flow rates and ratios, surfactants, and continuous phase viscosity. By using nanochannels with a height of 100–900 nm, 0.4–3.5 μm diameter droplets (volume down to 30 aL) have been produced. The generated droplets are stable for at least weeks.
Microfluidics has received enormous attention because of the ready availability of fabrication and measurement methods (Losey et al. 2002; Tien et al. 2002; Sinton 2004; van den Berg and Lammerink 1998). Many areas employ multiphase droplet-based microfludics (Shui et al. 2007; Gunther and Jensen 2006), including inkjet printers (van Dam and Le Clerc 2004), systems for separation of biochemical samples (Fujimura et al. 2003), manipulation of biomolecules (Lo et al. 2004), bio-sensing (Wang et al. 2006), single cell analysis (Yong et al. 2010), enhanced mixing for bio-sample reactions (Yang et al. 2006), biomolecular detection (Tseng et al. 2004), drug delivery devices (Chung et al. 2008), dairy analysis (Skurtys and Aguilera 2008), microelectronic cooling (Nilson et al. 2006), explosives detection (Piorek et al. 2007), bubble computing (Prakash and Gershenfeld 2007), interfacial tension measurement (Xu et al. 2008), and analysis of emulsions, foams, and bubble coalescence (Kralj et al. 2005). As general characteristics, the droplets should be as stable, monodisperse, reproducible, and controllable as possible.
Precise understanding of biological or chemical functions requires the ability to isolate and study single molecules. Although measurements can be made on single molecules in a dilute solution as they diffuse through the measurement volume, most such approaches rely on sophisticated optical strategies to limit the detection volume. Small containers can provide a means of confining single molecules inside of a measurement volume. We will need attoliter (or femtoliter) size containers to study single enzyme molecule activity with a μM (or nM) solution. The contents of individual containers should if possible be independently controllable so that distinct reactions can take place in separate containers. An emulsion in which one liquid forms droplets in another immiscible liquid (water-droplet-in-oil or oil-droplet-in-water) is a simple and attractive way of creating such containers (Lu et al. 1998; Nakano et al. 2003).
Microdroplets can be formed using flow-focusing, either by increasing shear gradients or by drawing the stream into a thin filament that breaks up by the Rayleigh-Plateau instability (Anna et al. 2003). Droplets generated in microfluidic devices are commonly created by squeezing or shearing the dispersed liquid (oil or water) into the continuous liquid (water or oil). Droplet size depends on microchannel geometry (Dollet et al. 2008; Sugiura et al. 2002b; Ravigururajan 1998; Link et al. 2004), flow rates (Garstecki et al. 2006; Martín-Banderas et al. 2005; Utada et al. 2007), and fluid properties (Ganan-Calvo and Riesco-Chueca 2006; Tice et al. 2004; Ganan-Calvo et al. 2004; Lam et al. 2002). Many parameters make the droplet formation process in such cases complex and sensitive to various environmental influences. Furthermore, the droplet size is limited by the microfabricated channel size (typically, in the micrometric range). Therefore, a method is required which can not only produce monodisperse attoliter droplets, but is also stable and robust, and inertial to small changes in the flow rate.
One of the methods to create droplets is by forming a thin liquid filament, which then spontaneously breaks up due to the capillary instability. The formation of a liquid filament is then a mandatory preliminary step for producing a droplet (Chen et al. 2003). This procedure is widely applied in flow-focusing devices. In these devices, an inner liquid thread can be created by outer liquid shearing (Ganan-Calvo and Gordillo 2001; Amyot and Plouraboue 2007; Garstecki et al. 2004; Gordillo et al. 2001; Barrero et al. 1998; Utada et al. 2007; Ganan-Calvo 1998). In this case, however, the confinement is not stable and can easily be disturbed by dynamic fluctuations.
A stable and strong confinement of a liquid filament, on the other hand, can be obtained by solid structures (nanochannels or nanoholes). Since this confinement stabilizes the liquid thread, breakup does not easily occur. When the confinement is subsequently removed by a sudden increase of channel diameter, capillary instability will cause formation of monodisperse droplets (van Dijke et al. 2010; Malloggi et al. 2010; van der Zwan et al. 2009, Sugiura et al. 2002a, b). Here, we investigated the droplet formation of this method and the scalability of this approach to produce droplets with a volume of less than 100 aL droplets (diameter below 600 nm) by using nanochannel sections down to 100 nm.
The devices were fabricated using standard photolithographic techniques. Nanochannels and microchannels were etched into a borosilicate glass wafer. The connection holes were drilled in a second borosilicate glass wafer using powder blasting techniques. Subsequently, these two wafers were aligned and thermally bonded together. We diced the bonded wafers to 10 × 20 mm-sized chips, which were mounted in a home-made chip holder and connected to gas-tight syringes (Microliter Syringes, Hamilton) via Nanoport connectors (Upchurch Scientific). Harvard syringe pumps (PHD 22/2000, Harvard Apparatus) were used to drive the liquid flow. The two-phase flow was visualized by an inverted microscope (Leica DMIRM) and recorded using a CCD camera (Orca ER).
The water phase was made fluorescent (appearing white in the microphotographs) by dissolving fluorescein sodium salt (Sigma–Aldrich Chemie GmbH, Germany) in de-ionized water at a final concentration of 0.01 mol L−1 (ηw = 1 mPa s). The organic phase (dark in the images) consisted of hexadecane (Sigma–Aldrich Chemie GmbH, Germany) without any added fluorescent markers (ηo = 3 mPa s). Glycerol (Sigma–Aldrich Chemie GmbH, Germany) has been used to tune the water phase viscosity from 1 to 10 mPa s (60 wt%). Surfactants have been used to stabilize generated droplets. The surfactants used were sodium dodecyl sulfate (CH3(CH2)11OSO3Na, SDS 99+%, Sigma–Aldrich Chemie GmbH, Germany), Tween80 (Sigma–Aldrich Chemie GmbH, Germany), and Span80 (Sigma–Aldrich Chemie GmbH, Germany). Solutions were prepared by dissolving hydrophilic surfactants (0.01 M SDS and 1 wt% Tween80) in water or hydrophobic surfactant (1 wt% Span80) in hexadecane. The surfactant concentrations were in all cases above the critical micelle concentrations. Solutions were degassed under vacuum for 1 h before using. All chemicals were used directly as received without further treatment. If it is not mentioned specifically, the two phases used are SDS (0.01 M) fluorescein (0.01 M) aqueous solution and hexadecane.
3 Results and discussion
3.1 Droplet formation at the nanochannel–microchannel interface
The liquid flow in the nanochannels is controlled using a syringe pump. The complications of conventional syringe pumping at ultralow flow rates (<10−2 μL min−1) were prevented by designing an integrated internal flow control system to create a flow split between a microchannel and the nanochannel, as previously reported by us and the group of Kitamori (Tamaki et al. 2006; Hibara et al. 2009; Shui et al. 2009a). The entire chip design is shown in Fig. 1. Liquids are introduced to the nanochannels (blue) from the sides by microchannels (red) which are connected between an inlet and a flow regulating channel. The total flow is divided between the nanochannel and the flow regulating microchannel, with a flow split ratio which is inversely proportional to their flow resistance ratio (Shui et al. 2009b). By designing the fluidic network, we could therefore control the flow in the nanochannel as required. We obtained flow rates of 10−5–10−3 μL min−1 in the nanochannel by using a typical syringe pump with a flow rate of 0.1–10 μL min−1 (split ratio on average 1:104).
Filaments occurring at different flow rates are shown in Fig. 4a–d. The length of the thinned filament decreases with increasing oil flow rate. Typically, we also found that the thin oil filaments were more prominent (longer and sharper) in shallower nanochannels. When a perturbation was applied to the tip, for example by a small obstruction in the constriction channel, the oil tip could split into two filaments (symmetric or asymmetric) close to the nanochannel exit (Fig. 4e–h). Same sized droplets were observed to break up from these two oil filaments.
3.2 The dependence of the droplet size on the nanochannel height
When an unconfined steady liquid filament (thread) spontaneously breaks up, the droplet size is proportional to the filament size in a certain range of experimental parameters (Ganan-Calvo and Gordillo 2001; Ganan-Calvo 1998; Utada et al. 2007). For the breakup of confined threads on expansion in a larger channel, it has been found in micro-devices that the droplet size is proportional to the dimension of the confining structure (van Dijke et al. 2010; Malloggi et al. 2010; van der Zwan et al. 2009). Here we created nanoconfinements of the oil threads by employing devices with nanochannel heights h of 100, 160, 260, 520, and 900 nm; widths w of 10, 20, and 50 μm; and lengths l of 500 and 1000 μm. The droplet diameter was optically measured in the microchannel section where the droplets are spherical.
We have also investigated the influence of the water phase viscosity (in the range of 1–10 mPa s by addition of glycerol) and of the use of different surfactants (SDS, Tween80, and Span80), resulting in a surface tension in the range of 10−4–10−2 N m−1) (Shui et al. 2009c) on the droplet formation process, to find out whether this method is sensitive to slight changes in fluidic properties. We did not observe a significant change of droplet size with water phase viscosity and nature of surfactants in the platform regime. However, the transition values from the platform region to the flow-rate dependent region were influenced by the relative liquid viscosities and surfactants. The transition occurs at lower flow rates if the water (continuous phase) viscosity increases. When using different surfactants, the transition occurred at lower flow rate in the order Span80 > Tween80 > SDS. Typically, the frequency of droplet formation in the devices ranged from 10 to 104 Hz, dependent on the flow rate. In future, the droplet production rate could be increased by using parallel nanochannels.
4 Conclusions and outlook
We have demonstrated the production of monodisperse attoliter droplets using nano-microfluidic devices. Over a wide range of absolute and relative flow rates, highly stable and monodisperse small (0.4–3.5 μm diameter) droplets were obtained in devices with nanochannel heights of 100–900 nm. The shallower nanochannels furthermore demonstrated wider flow rate operating ranges. The droplet size was linearly related to the nanochannel height. In this range, we did not observe obvious effect of liquid flow rates, surfactants and continuous phase viscosity on generated droplet size. Thus, we can predict droplet size generated in an existing device or design such a device to create droplets as required. Limitations to the obtainable droplet size will be posed by nanofabrication technology. The attoliter droplets can be used as small containers which will on average encapsulate one molecule per droplet at analyte concentrations in the μM range, forming a promising tool for single molecular studies. By using this method, we can further scale-down the droplet size by using shallower nanochannels, and scale-up the droplet production by using parallel nanochannels.
This research was supported by the Dutch Ministry of Economic Affairs through a Nanoimpuls grant. The authors gratefully acknowledge Wim van Hoeve and Chao Sun for supporting with high speed camera.
This article is distributed under the terms of the Creative Commons Attribution Noncommercial License which permits any noncommercial use, distribution, and reproduction in any medium, provided the original author(s) and source are credited.