Utility of Low-Cost, Miniaturized Peristaltic and Venturi Pumps in Droplet Microfluidics

Abstract

Many laboratory applications utilizing droplet microfluidics rely on precision syringe pumps for flow generation. In this study, the use of an open-source peristaltic pump primarily composed of 3D printed parts and a low-cost commercial Venturi pump are explored for their use as an alternative to syringe pumps for droplet microfluidics. Both devices provided stable flow (<2% RSD) over a range of 1 – 7 μL/min and high reproducibility in signal intensity at a droplet generation rate around 0.25 Hz (<2% RSD), which are comparable in performance to similar measurements on standard syringe pumps. As a novel flow generation source for microfluidic applications, the use of the miniaturized Venturi pump was also applied to droplet signal monitoring studies used to measure changes in concentration over time, with average signal reproducibility <4% RSD for both single-stream fluorometric and reagent addition colorimetric applications. These low-cost flow methods provide stable flow sufficient for common droplet microfluidic approaches and can be implemented in a wide variety of simple, and potentially portable, analytical measurement devices.

1. Introduction

Segmented flow (or “droplet”) microfluidic techniques have found utility in a broad range of application areas in chemical analysis, including high-throughput screening, cell culture monitoring, clinical diagnostics, synthetic reaction monitoring, and single-cell analysis [1–7]. Droplets are especially useful when high temporal resolution sampling of biological systems is desired [8–15]. Often, droplets are generated in microfluidic systems utilizing syringe pumps for flow generation [16,17]. However, the cost and size of typical syringe pumps limit the ability to significantly miniaturize an entire analytical system utilizing droplet microfluidics in a cost-effective way [18]. Electroosmotic pumping and capillary flow action are more difficult to implement with segmented flow, so alternative approaches to pressure-driven flow beyond syringe pumps provide the best opportunity for low-cost systems that utilize droplet streams.

One option is the use of peristaltic pumping, which is a very common approach to generating flow in low-pressure systems. Peristaltic pumping, despite its well-known flow rate oscillations due to the peristalsis mechanism, can be designed for easily tunable, sufficiently precise flow. In microfluidic devices, two key approaches to peristalsis are often utilized [17]: (1) coupling device inlets or outlets to expensive commercial peristaltic pumps capable of low μL/min flow rates, or (2) integrating peristaltic pumps within/around the microfluidic device by treating device channels as the tubing that is compressed to generate peristalsis [19–24]. As the latter option can be difficult to fabricate and control with low-cost microfabrication approaches, our goal here was to design a standalone open-source peristaltic pumping system that is both simple and inexpensive. A few open-source peristaltic pumps have been previously reported in the literature [25–28], but these are not well-suited for the low flow rates required for microfluidic device operation, with the lowest reported values only in the 10 – 100 μL/min range [28]. Here, we describe the fabrication and use of a low-cost, open-source peristaltic pump using parts that are either 3D printed or commercially available and inexpensive, designed specifically to achieve sub-10 μL/min flow rates suitable for droplet microfluidics.

The other common approach for low-cost flow generation is the use of the Venturi effect in which the flow of gas or liquid can be used to generate a partial vacuum without moving parts. When a stream of one of these fluids passes through a narrow channel before exiting a wider output channel, a pressure drop occurs that effectively generates a partial vacuum that can be used to pull fluids through a microfluidic device. The use of Venturi pumping in microfluidic devices has generally focused on the fabrication of this channel restriction design directly within the microfluidic device [29–33], although this can require more complex device preparation strategies. Droplet generation with Venturi flow in tubing has also been used for well-plate sampling coupled to easy ambient sonic-spray ionization mass spectrometry [34]. Here, a commercially available miniaturized Venturi pump is investigated as a source of vacuum for driving flow in microfluidic devices, an approach that has not yet been demonstrated to date to the best knowledge of the authors. This external pump can be re-used across many microfluidic devices, simplifying flow generation for microfluidic devices at a greatly reduced price compared to syringe pumps. Additional benefits of this type of Venturi pump include its portability (both in terms of size and lack of power requirement), low-cost, and robustness (no moving parts).

In this study, we evaluate the use of an open-source peristaltic pump (Figures 1A and 1B, cost of approximately $80), a miniaturized Venturi pump (Figure 1C, cost of approximately $100), and a standard syringe pump (cost of $1,650) in withdraw mode for generating droplet flow from the outlet of a microfluidic device. The precision and reproducibility of droplet formation using all three methods were compared on chip formats that are typically utilized when sampling from biological or environmental systems. The miniaturized Venturi pump was further explored for its use in microfluidic devices designed for on-line sampling of aqueous systems using droplets and droplet reagent addition, typical functions in droplet-based chemical analysis.

Figure 1:

Images of the peristaltic pump (side in A, top-down in B) and miniaturized Venturi pump (C).

2. Materials and Methods

2.1. Reagents

Perfluorodecalin (PFD) oil, trichloro(1H,1H,2H,2H-perfluorooctyl)silane (“fluorosilane”), and iron (III) chloride were purchased from Alfa Aesar (Tewksbury, MA). Resorufin was obtained from Marker Gene Technologies (Eugene, OR). Potassium thiocyanate was purchased from Millipore Sigma (St. Louis, MO). Isopropanol (IPA, reagent grade) was received from VWR International (Bridgeport, NJ). Deionized water (18.2 MΩ) was obtained from a Barnstead GenPure UV-TOC/UF xCAD Plus purifier (Thermo Electron, Germany).

2.2. Microfluidic Device Fabrication

Three-dimensional master molds for polydimethylsiloxane (PDMS) soft lithography fabrication were designed in Autodesk Inventor Professional 2018 (San Rafael, CA). The mold designs (Figure S1), which consisted of a flat base structure and raised inverted channels, were exported as .STL files and printed using a Miicraft+ Digital Light Processing Stereolithography (DLP-SLA) 3D printer with proprietary microfluidic resin BV-007 (Miicraft, Hsinchu, Taiwan). The .STL files were sliced with 50 μm thickness using the Miicraft MiiUtility software, and each layer was cured for 20 s with a 405 nm light-emitting diode (LED). After the print was completed, the mold was placed in an isopropanol bath for 5 min, then dried and placed in the Miicraft+ UV cure chamber for 10 min. Following UV curing, the molds are placed in an IPA bath for 4 hr, dried, and then suspended 1.5 cm above an aliquot of 10 μL of the fluorosilane in a vacuum oven for 3 hr at 40 °C as a surface treatment to improve the PDMS peel-off process, based on previously published procedures for mold fluorination [35,36]. A layer of PDMS (10:1 base:curing agent, Sylgard 184, Dow Corning, Midland, MI) approximately 2 mm thick was poured over the fluorinated mold and cured for 3 hours at 40 °C. This temperature is slightly lower than typical PDMS curing (60 – 80 °C) to avoid minor warping of the thin, 3D printed mold that can affect channel structure. The PDMS with imprinted channels was removed from the mold and then bonded to a blank cured piece of PDMS following 5 min of plasma oxidation with a corona air plasma treater (BD-20AC, Electro-Technic Products, Chicago, IL). The bonded chip was then placed in an oven at 40 °C for 1 hr. Access to the enclosed channels following bonding was achieved by cutting the PDMS at each channel end, and then inserting 360 μm outer diameter (o.d.) fused silica tubing of varying inner diameter (i.d.) (Polymicro Technologies, Phoenix, AZ). Once capillaries were integrated into the device (Table S1), an additional layer of PDMS was poured around the device to eliminate leakages from the chip. Chips were typically used for ~4 weeks, after which analyte adhesion to channel walls was observed.

2.3. Flow Control Techniques and Calibration

As a control technique for flow generation, a Chemyx Fusion 200 syringe pump (Stafford, TX) equipped with a 100 μL gas-tight syringe (Hamilton, Franklin, MA) was used in withdraw mode. The syringe was connected to the fused silica capillary at the chip outlet (Figure S1) using a 1/16” PEEK union with the blunt needle end of the syringe sheathed in .030” i.d. PEEK tubing and the fused silica capillary sheathed in a .0155” i.d. NanoTight sleeve (all connectors from Idex, Oak Harbor, WA). A low-cost, open-source peristaltic pump was also designed for this study and was used for flow generation by connecting it to the outlet of the fluidic system. Details on its fabrication and user control can be found in Appendix A of the Supporting Information. Finally, a miniaturized Venturi vacuum pump (part EW-78165–00) from Cole-Parmer (Vernon Hills, IL) was tested as a second low-cost flow generation approach, with flow rate control achieved by varying the gauge pressure of a compressed air tank. Fluidic outlets were connected to a small vacuum trap, which was then connected to the Venturi pump (Figure S2). To correlate user-controlled variables with flow rate (programmed rate for the syringe pump, motor rotation speed for the peristaltic pump, and tank gauge pressure for the Venturi pump), the outlet of a PDMS microfluidic tee designed for generating droplet flow (Table S1, Chip A) was coupled to a liquid flow meter (SLG-0150, Sensirion, Chicago, IL) and the flow rate of deionized water from two sample reservoirs upstream from the tee was recorded at a 25 Hz sampling rate. In this study, the Venturi pump and peristaltic pump were implemented in a “pull” mode in which flow is generated by drawing liquid toward the device outlet from reservoirs and flow rates in individual channels are based on their comparative flow resistance [37,38]. To identify the vacuum pressure generated by the miniaturized Venturi pump at various gauge pressures, it was coupled directly to a GPS-BTA gas pressure sensor on a LabQuest2 sensor interface (Vernier, Beaverton, OR) (Figure S3). As an equivalent comparison, the syringe pump was operated in withdraw mode. Calibration curves comparing these user-controlled variables with the measured flow rates are shown in Figure S4.

2.4. Segmented Flow Droplet Generation and Measurement

Droplet generation was compared for all three flow methods in terms of formation rate and reproducibility of signal intensity. Fluorescence detection was used to monitor droplets containing 50 nM resorufin that were generated using Chip B (Figure S1) with an Olympus IX-51 epifluorescence microscope (Center Valley, PA) coupled to a photomultiplier tube (PMT, H10721–01, Hamamatsu Photonics, Bridgewater, NJ). Excitation was generated from an arc lamp (LH100HG, Olympus) with a 530–550 nm excitation/590 nm emission filter cube (U-MNG2, Olympus). The PMT output was filtered (10 Hz, 12 dB lowpass) and amplified on a SR570 current preamplifier (Stanford Research Systems, Sunnyvale, CA) and recorded on a LabView program (National Instruments, Austin, TX) at a 21 Hz acquisition rate. The same epifluorescence system was used to measure the signal response change following spike additions (100 μL of a 25 μM resorufin solution) to a stirred vial that initially contained 1 μM resorufin as a demonstration of the applicability of this approach for sampling real-time concentration changes in biological systems. For this experiment, a 23 cm segment of 75 μm i.d. fused silica capillary was placed into the stirred vial and connected to a 4 cm segment of 100 μm i.d. fused silica capillary that had been placed in a PFD reservoir using a tee with 150 μm i.d. channels (C360QTPK6, Valco Instruments Co. Inc., Houston, TX). The outlet of the tee was connected to the Venturi pump with a 16 cm segment of 150 μm i.d. tubing and signal was detected at a window in the tubing located 5 cm from the tee. Reagent addition capabilities were examined using Chip C (Figure S1) and the Venturi pump, with a stereo microscope (1–4x magnification) coupled to a Moticam 1080 HD camera (National Optical & Scientific Instruments, Inc., Schertz, TX). Signal was recorded as RGB intensity over time with ImageJ (National Institutes of Health, Bethesda, MD) [39] and the “intensity vs. time” function in the “ImageJ for Microscopy” plug-in (McMaster Biophotonics Facility, Hamilton, ON) [40]. Acquired data was analyzed and plotted using Microsoft Excel (Redmond, WA) and Igor Pro 6.0 (Wavemetrics, Inc., Lake Oswego, OR).

3. Results and Discussion

3.1. Comparison of Flow Methods for Droplet Generation

In this work, a commercial, miniaturized Venturi vacuum pump and an open-source design for peristaltic pumping were explored for use as low-cost approaches for segmented flow droplet generation in microfluidic devices. These new techniques were compared to a syringe pump, a more standard approach for this purpose, in terms of flow rate control using a liquid flow meter (Figure S4). As shown in Figure S4A, the programmed syringe pump flow rate and the flow rate recorded by the flow meter were effectively identical (y = 0.9668x + 0.0226, R² = 0.9999) over the range of 1 – 7 μL/min, a common range for microfluidic flow generation in channels of these dimensions. A similar flow range was achieved for the peristaltic pump when operated at a rotation per minute (RPM) speed between 3 and 25, with a linearity of R² = 0.9971 (Figure S4B). For the miniaturized Venturi pump, these flow rates were achieved with applied gas tank gauge pressures of 20 – 50 psi, with R² = 0.9928 (Figure S4C). Over a time period of 60 s in which the flow meter reading was recorded at 25 Hz, the average %RSD values across the entire flow ranges were 0.42%, 1.13%, and 1.81% for the syringe pump, peristaltic pump, and Venturi pump, respectively. Based on these results, the peristaltic pump and miniaturized Venturi pump perform slightly below the syringe pump in terms of linearity and reproducibility, but are still sufficient for most microfluidic applications in which stable flow is needed, providing suitable alternatives for use in low-cost system designs.

The comparison in the preceding paragraph focused on flow rates utilizing water as a solvent from both chip inlets based on compatibility with the programmed flow meter calibration. In segmented flow droplet microfluidics, in which the flow resistance can change based on the differences in viscosity between the oil and aqueous phases as well as relative channel dimensions, characterizing performance based on droplet generation can be more appropriate. Here, a comparison between the three flow modes was made using fluorescent intensity measurements of segmented aqueous fluorescent dye droplets. This is a common detection mode for droplet microfluidics and can provide additional quantitative comparisons of droplet intensity between the three flow techniques. Figure 2 shows droplet streams for the syringe pump at 5.5 μL/min, the peristaltic pump at 15 RPM, and the Venturi pump at a gauge pressure 40 psi. In each of these cases, the droplet formation rate for each approach was 0.23 Hz (average volume of 146 nL), 0.25 Hz (average volume of 108 nL), and 0.25 Hz (average volume of 135 nL), with %RSD values for droplet intensity of 2.15%, 1.36%, and 1.54%, respectively. The peristaltic pump does demonstrate slight variability in droplet width due to the slight flow fluctuations from the peristalsis mechanism, but this does not affect the measured intensity that is the key factor for quantitative measurements. Overall, these results indicate that these low-cost approaches to flow generation for segmented flow microfluidics compare favorably with standard syringe pump approaches under these conditions at droplet similar generation rates to other online droplet detection approaches (0.1 – 0.4 Hz in [13,41,42]). Because of the maximum pressure that can be used with the Venturi pump (80 psi gauge pressure) and issues at high rotational speeds with the peristaltic pump (heat generation, flow fluctuations, and reduced torque), syringe pumps may still be more useful for faster droplet generation at higher flow rates, albeit at a higher cost.

Figure 2:

Comparison of droplet formation rate and signal intensity for three speeds using each flow generation technique measured for a 50 nM resorufin solution on an epifluorescence microscope. Droplet streams for a syringe pump flow rate of 5.5 μL/min (red trace), a peristaltic pump motor speed of 15 RPM (black trace), and a Venturi pump tank gauge pressure of 40 psi (blue trace) are shown. Droplet intensity values were normalized to the average signal for a given measurement set, with values offset by 0.1 units for graphical depiction.

3.2. Droplet Microfluidic Applications using Miniaturized Venturi Pumping

In chemical analysis, one key application of droplet microfluidics is the ability to sample a transient system with high temporal resolution, and maintain that resolution from the sampling point to the detection point by minimizing diffusion-based broadening of the analyte plug [8]. For low-cost environmental monitoring applications using droplets, such as tracking chemical species in flowing water sources [41–43], the potential to utilize a “power-free” flow generation technique such as the miniaturized Venturi pump can provide unique opportunities for long-term autonomous analysis [44]. By utilizing a small apparatus that contains no moving parts and only requires a compressed air source, the capability to conduct on-site analysis implementing droplet microfluidics without mechanical failure is enhanced, making the Venturi approach a promising alternative to a syringe pump or even the low-cost peristaltic pump. With its demonstrated capability to reproducibly generate segmented flow droplets, we implemented the miniaturized Venturi pump into two widely used analytical methods of monitoring samples with a transient signal.

First, a similar experimental set-up that was used for Figure 2 was adapted to draw aqueous sample from a 1 mL stirred vial containing 1 μM resorufin dye, selected for its use in many fluorescence-based biological assays. In Figure 3, the results of four subsequent spikes (100 μL of 25 μM resorufin) spaced at 4 min intervals are shown (Figure 3A for all spikes and Figure 3B for closer view of first spike), along with a calibration curve based on peak fluorescent intensity from each droplet (Figure 3C). Across the range of 1 – 9.8 μM resorufin, the curve had R² = 0.9929 and an average droplet intensity RSD value across all the concentrations of 3.9%. This average RSD indicates that droplet signal was stable once the concentration at each level had been fully mixed in the stirred vial. The average rise time for each step change was 27.4 s (3.3% RSD), which means that changes in analyte signal that occur on this time scale or longer can successfully be monitored with this chip design and Venturi-based flow approach.

Figure 3:

Demonstration of droplet monitoring over time with multiple reagent spikes. An initial 1 mL stirred vial of 1 μM resorufin was spiked with four 100 μL aliquots of 25 μM four minutes apart as shown with the black indicator arrows (A). A magnification of the signal increase after the first spike is shown in (B) and a calibration curve for all five concentration regions is shown in (C). The equation of the calibration curve is y = 0.0377x + 0.0176 with R² = 0.9929.

A second microfluidic device design with a Y-junction was then used to characterize the reagent addition in droplets using the Venturi flow approach. Droplet flow provides enhanced mixing of solutions based on the small volume of droplets and the convective flow that is created as they move in a segmented stream [45], providing a critical tool for assays that require additional reagents to generate a detectable product [13,46–48]. Here, the formation of the red thiocyanatoiron complex was used for the colorimetric measurement (by RGB value characterized with ImageJ) of the reaction between potassium thiocyanate and iron (III) chloride as an example of monitoring iron content in water. The U.S. Environmental Protection Agency has identified a series of secondary drinking water contaminants that can affect water taste, color, and/or odor [49]. The concentration threshold for iron on this list is 0.3 mg/L. Initially, an inlet channel containing stream water collected from Rowan University was combined with a second inlet channel containing potassium thiocyanate at a segmentation junction (chip ‘C’ in Figure S1). The changes in signal over time as two aliquots of iron (III) chloride were added to the water reservoir are shown in Figure 4 (visualization is shown in Figure S5). The spikes provided added iron concentrations in the sample of 0.2 mg/L (below the EPA threshold) and 0.4 mg/L (above the EPA threshold). This initial stream provided a small noise signal due a minimal level of refraction at the aqueous-oil interface, but once the Fe³⁺ was added, a step change was immediately observed within two droplets (approximately 9 s), demonstrating a faster rise time as droplet formation occurred closer to the sampling point [9]. The average droplet volume was 160 nL (3.2% RSD), which is slightly larger than the ~100 nL volume of one recent study [13] but an order of magnitude smaller (1.5 – 4 μL volume) than other downstream applications [41,42]. The average RSD value for droplet intensity across the two detection levels was 2.8% at a droplet formation rate of 0.23 Hz. The results of Figures 3 and 4 demonstrate the capability of this new miniaturized Venturi pump approach to provide stable flow for droplet microfluidic applications used to monitor transient chemical concentrations in aqueous systems, including the potential for tracking the performance of iron remediation strategies in water treatment facilities using the colorimetric test demonstrated here.

Figure 4:

Formation of red thiocyanatoiron complex using reagent addition approach. A 5 μL and a 8.5 μL spikes of a 1 mg/L solution of FeCl₃ were added to 20 μL of stream water in a stirred vial, which served as one inlet stream connected to additional inlet streams of 5 mg/L potassium thiocyanate and PFD. Signal was measured as the RGB intensity value in ImageJ.

4. Conclusions

In this study, two low-cost approaches for the generation of low μL/min flow rates in microfluidic devices were demonstrated for the use of segmented flow droplet generation and compared to the more widely used strategy of utilizing commercial syringe pumps. Both the open-source peristaltic pump and the miniaturized low-cost Venturi pump demonstrated stable flow and low RSD values in droplet formation. This Venturi flow approach was also successfully adapted for use in droplet-based monitoring of aqueous sample streams with changing concentration, either directly or through the use of reagent addition to generate a detectable product.

The use of low-cost approaches to generate flow in microfluidic devices will continue to grow in importance as more remote and/or point-of-care platforms are needed for on-site chemical analysis. These new techniques are especially advantageous as they either have low power requirements (peristaltic pump) or can be operated completely independent of a power source (miniaturized Venturi pump using compressed air). In the future, these new droplet formation strategies will be applied for specific applications in environmental monitoring that can benefit from these advantages.

Highlights.

• Low-cost flow approaches were designed for segmented flow droplet microfluidics

• Open-source peristaltic pump constructed primarily with 3D printing

• Novel use of miniaturized Venturi pump used for microfluidic flow

• Flow generation and droplet formation is comparable to syringe pumps

• Chemical measurements using droplets with Venturi approach were achieved