The protocol demonstrates a convenient method to produce harmonic oscillatory flow from 10-1000 Hz in microchannels. This is performed by interfacing a computer-controlled speaker diaphragm to the microchannel in a modular manner.
Microfluidic technology has become a standard tool in chemical and biological laboratories for both analysis and synthesis. The injection of liquid samples, such as chemical reagents and cell cultures, is predominantly accomplished through steady flows that are typically driven by syringe pumps, gravity, or capillary forces. The use of complementary oscillatory flows is seldom considered in applications despite its numerous advantages as recently demonstrated in the literature. The significant technical barrier to the implementation of oscillatory flows in microchannels is likely responsible for the lack of its widespread adoption. Advanced commercial syringe pumps that can produce oscillatory flow, are often more expensive and only work for frequencies less than 1 Hz. Here, the assembly and operation of a low-cost, plug-and-play type speaker-based apparatus that generates oscillatory flow in microchannels is demonstrated. High-fidelity harmonic oscillatory flows with frequencies ranging from 10-1000 Hz can be achieved along with independent amplitude control. Amplitudes ranging from 10-600 µm can be achieved throughout the entire range of operation, including amplitudes > 1 mm at the resonant frequency, in a typical microchannel. Although the oscillation frequency is determined by the speaker, we illustrate that the oscillation amplitude is sensitive to fluid properties and channel geometry. Specifically, the oscillation amplitude decreases with increasing channel circuit length and liquid viscosity, and in contrast, the amplitude increases with increasing speaker tube thickness and length. Additionally, the apparatus requires no prior features to be designed on the microchannel and is easily detachable. It can be used simultaneously with a steady flow created by a syringe pump to generate pulsatile flows.
The precise control of liquid flow rate in microchannels is crucial for lab-on-a-chip applications such as droplet production and encapsulation1, mixing2,3, and the sorting and manipulation of suspended particles4,5,6,7. The predominantly used method for flow control is a syringe pump that produces highly controlled steady flows dispensing either a fixed volume of liquid or a fixed volumetric flow rate, often limited to entirely unidirectional flow. Alternative strategies for producing unidirectional flow include using gravitational head8, capillary forces9, or electro-osmotic flow10. Programmable syringe pumps allow for a time-dependent bidirectional control of flow rates and dispensed volumes but are limited to response times greater than 1 s due to the mechanical inertia of the syringe pump.
Flow control at shorter time scales unlocks a plethora6,11,12,13,14,15 of otherwise inaccessible possibilities due to qualitative changes in flow physics. The most practical means of harnessing this varied flow physics is through acoustic waves or oscillatory flows with time periods ranging from 10-1– 10-9 s or 101 -109 Hz. The higher end of this frequency range is accessed using bulk acoustic wave (BAW; 100 kHz-10 MHz) and surface acoustic wave (SAW; 10 MHz-1 GHz) devices. In a typical BAW device, the entire substrate and the fluid column are vibrated by applying a voltage signal across a bonded piezoelectric. This enables relatively high throughputs but also results in heating at higher amplitudes. In SAW devices, however, the solid-liquid interface is oscillated by applying voltage to a pair of interdigitated electrodes patterned on a piezoelectric substrate. Due to the very short wavelengths (1 µm-100 µm) particles as small as 300 nm can be precisely manipulated by the pressure wave generated in SAW devices. Despite the ability to manipulate small particles, SAW methods are limited to local particle manipulation since the wave rapidly attenuates with distance from the source.
At the 1-100 kHz frequency range, oscillatory flows are usually generated using piezo-elements that are bonded to a polydimethylsiloxane (PDMS) microchannel above a designed cavity16,17. The PDMS membrane above the patterned cavity behaves like a vibrating membrane or drum that pressurizes the fluid within the channel. At this frequency range, the wavelength is larger than the channel size, but the oscillation velocity amplitudes are small. The most useful phenomenon in this frequency regime is the generation of acoustic/viscous streaming flows, which are rectified steady flows caused due to non-linearity inherent in the flow of liquids with inertia18. The steady streaming flows typically manifest as high-speed counter-rotating vortices in the vicinity of obstacles, sharp corners, or micro-bubbles. These vortices are useful for mixing19,20 and separating 10 µm sized particles from the flow stream21.
For frequencies in the range of 10-1000 Hz, both the velocity of the oscillatory component and its associated steady viscous streaming are considerable in magnitude and useful. Strong oscillatory flows in this frequency range can be used for inertial focusing22, facilitate droplet generation23, and can generate flow conditions (Womersley numbers) that mimic blood flow for in vitro studies. On the other hand, streaming flows are useful for mixing, particle trapping, and manipulation. Oscillatory flow in this range of frequencies can also be accomplished using a piezo-element bonded to the device as described above23. A significant hurdle to implementing oscillatory flows through a bonded piezo element is that it requires features to be designed beforehand. Furthermore, the bonded speaker elements are not detachable, and a new element must be bonded to each device24. However, such devices present the advantage of being compact. An alternative method is using an electromechanical relay valve20. These valves require pneumatic pressure sources and custom control software for operation and therefore increase the technical barrier to testing and implementation. Nevertheless, such devices enable the application of set pressure amplitude and frequency.
In this article, the construction, operation, and characterization of a user-friendly method to generate oscillatory flows in the frequency range of 10-1000 Hz in microchannels is described. The method offers numerous advantages such as cost-effective assembly, ease of operation, and ready to interface with standard microfluidic channels and accessories such as syringe pumps and tubing. Additionally, compared to previous similar approaches25, the method offers the user selective and independent control of oscillation frequencies and amplitudes, including the modulation between sinusoidal and non-sinusoidal waveforms. These features allow users to easily deploy oscillatory flows and, therefore, facilitate widespread adoption into a broad range of currently existing microfluidic technologies and applications in the fields of biology and chemistry.
1. Rapid prototype mold design and fabrication
2. PDMS microchannel fabrication
3. Oscillatory driver assembly
4. Adapter assembly
NOTE: The complete speaker-to-tube adapter assembly is illustrated by the schematic in Figure 1.
5. Operation of the experimental setup for oscillatory flows in microchannels
6. Observation and amplitude measurement
To illustrate the capability and performance of the above setup, representative results of oscillatory flow in a simple linear microchannel with a square cross-section are presented. The width and height of the channel are 110 µm and its length is 5 cm. First, we describe the motion of spherical polystyrene tracer particles and how these can be used to check the fidelity of the oscillatory signal as well as the range of oscillation amplitudes achievable. We then discuss the effect of specific fluid properties or microfluidic materials on oscillation amplitude. Finally, we illustrate the capability for non-sinusoidal waveforms.
For comparison, we define the reference case by the following fluid properties, channel geometry, and microfluidic materials. The working liquid is deionized water (µ = 1.00 mPa.s) with 0.01% volume fraction of tracer particles which have diameter, d = 1 µm and density, ρ = 1.20 kg/m3. The corresponding particle response time, given by ρd2/18µ, is 70 ns which is far less than the corresponding oscillatory time scales (1-100 ms). The particles are observed at the channel mid-height with a 10x objective and a depth of focus of 10 µm. The microfluidic tube has diameters 1.27 mm x 0.76 mm (outer x inner) and an outlet tube length of 12 cm that is held 5 cm above the channel level.
The tracked displacements of tracer particles at the channel midplane for different oscillation frequencies are shown in Figure 2. A harmonic signal is observed for all of the oscillation frequencies shown, which are 100 Hz, 200 Hz, 400 Hz, and 800 Hz. The imaging frame rate was greater than or equal to 20 times the oscillation frequency. The amplitude (speaker volume) setting was maintained constant across the different oscillation frequencies. For the frequencies 100 Hz, 200 Hz, 400 Hz, and 800 Hz, the corresponding amplitudes are approximately 125 µm, 100 µm, 25 µm, and 10 µm, respectively.
The tracked displacement of particles is also used to determine the fidelity of the harmonic motion and the range of oscillation amplitudes, a critical step in the calibration process. The fidelity of the harmonic displacement of particles at different oscillation frequencies and amplitudes is illustrated using the Fourier spectra and shown in Figure 3A. For frequencies of 50 Hz, 200 Hz and 400 Hz respectively, three different amplitudes characterized by the potential difference in the aux cable (or amplifier input voltage) are considered. The settings are named low (30%, 1.5 V, yellow), intermediate (60%, 3 V, orange), and high (90%, 4.5 V, red). Here, the percentage represents the magnitude of the volume setting with respect to the maximum speaker volume, or corresponding voltage of 5 V. The Fourier spectra of particle displacement at oscillation frequencies of 50 Hz, 200 Hz, and 800 Hz are shown in Figure 3A for three different amplifier input voltages (1.5 V, 3 V, 4.5 V) corresponding to yellow, orange and red colors respectively. The primary peak of the spectrum corresponds exactly to the applied frequency for all volume settings. The primary peak is > 10 times the secondary peaks, even at the highest amplitude.
For an amplifier input voltage of 5 V, the amplitude of the speaker cone displacement has a maximum value of 5 mm and remains a constant for frequencies up to 50 Hz and then decreases approximately quadratically for frequencies above 50 Hz (e.g., 1.5 mm at 100 Hz). The particle oscillation amplitude in the liquid is proportional to the power transduced given by the product of the speaker cone amplitude and the oscillation frequency. We therefore expect that the oscillatory amplitude is maximum near the speaker resonant frequency and decreases for frequencies on either side of it for a fixed amplifier input voltage. Further, we may also expect that the oscillatory amplitude of the fluid varies linearly with the amplifier input voltage and its value cannot exceed that of the speaker cone amplitude.
These expectations are confirmed in a plot of oscillation amplitude versus frequency shown in Figure 3B. For all speaker volume settings, the characteristic curve has a resonant peak, which occurs at approximately 180 Hz, beyond which the amplitude decreases with increasing frequency. The curves at different voltages appear identical except for vertical translations in log-scale implying that the oscillatory amplitude varies linearly with voltage. Finally, the maximum amplitude is less than 1.5 mm even at the resonant frequency of 5 V. Nevertheless, a volume setting can be selected such that oscillation amplitudes of > 100 µm can be achieved over the entire operational frequency range.
Next, select example cases are presented on the effect of the liquid viscosity, the tube diameter, and tube length on the oscillatory amplitude over the range of operational frequencies with respect to the reference case described above. For these experiments, the driver amplitude (speaker volume) is maintained constant at the intermediate level and only one setup parameter is modified at a time while the remaining parameters are identical to the reference control case (diamond symbols). The corresponding results for oscillation amplitude versus frequency are shown in Figure 4. When the viscosity of the working liquid is increased by changing to a 25% glycerol solution (µ = 1.81 mPa.s) the amplitude decreases by a factor of nearly 2 over the range of operating frequencies (square symbols). This suggests that, in general, increasing the liquid viscosity compared to that of deionized water would result in a similar characteristic amplitude versus frequency curve with a constant factor decrease in the amplitude. When the microfluidic tubing diameter for the same material (polyethylene) is increased to 2.41 mm x 1.67 mm, the amplitude increases compared to the reference case by a factor between 1.5-3 depending on the frequency (circle symbols). The increase is larger at high frequencies and smaller at low frequencies, indicating the resonant frequency has increased. When the tube length for the same material (polyethylene) is increased to 24 cm (by a factor of 2), the amplitude increases significantly near the resonant frequency but remains unchanged from the reference control case at very low and very high frequencies (triangle symbols).
In addition to the sinusoidal waveforms discussed above, non-sinusoidal waveforms are also demonstrated. Particle displacement tracks for square, triangle, and sawtooth waveforms are shown in Figure 5A. Here, the amplitude setting is intermediate (60% of maximum), the driving frequency is 100 Hz, and particles are observed at 4000 frames/s. As expected, very sharp changes in position associated with square and sawtooth waveforms are not possible in real systems with a finite response time. For this speaker system, the response time may be estimated to be 0.5 ms. Nonetheless, the Fourier spectra of these waveforms are observed to be in good agreement with the ideal spectra, at least up to the third harmonic as shown in Figure 5B.
Figure 1. A schematic to illustrate the apparatus design and assembly. The critical components are (I) speaker, (II) speaker mount, (III) speaker-to-tube adapter, (IV) pipette-tip wedge seal, (V) polyethylene tubing, and (VI) PDMS microchannel. Please click here to view a larger version of this figure.
Figure 2. Examples of particle displacement during oscillatory flow. Representative particle tracks during sinusoidal waveform input at different frequencies were obtained using high-speed imaging. Please click here to view a larger version of this figure.
Figure 3. Analysis of particle displacement for signal fidelity and amplitude range. (A) Fourier spectrum analysis of sinusoidal oscillations at different oscillation frequencies and amplitudes, or speaker volumes. (B) The characteristic curve of the oscillation amplitude versus frequency at three different speaker volume settings. Please click here to view a larger version of this figure.
Figure 4. Effects of tube length, tube diameter, and liquid viscosity on oscillatory amplitude. When compared to the reference case, an increase in tube length or tube diameter will lead to an increase in oscillation amplitude over the range of operational frequencies. An increase in viscosity, however, decreases the oscillation amplitude. Please click here to view a larger version of this figure.
Figure 5. Examples of non-sinusoidal waveforms. (A) Particle displacements for square, triangular, and sawtooth waveforms at an oscillation frequency of 100 Hz. (B) The corresponding Fourier spectra for non-sinusoidal particle displacements. Please click here to view a larger version of this figure.
Supplementary File 1. Stereolithography file to produce a 3D printed speaker mount referred to in Figure 1 (II). Please click here to download this File.
Supplementary File 2. Stereolithography file to produce a 3D printed speaker tube adapter referred to in Figure 1 (III). Please click here to download this File.
We have demonstrated the assembly (see protocol critical steps 3 and 4) and operation (see protocol critical steps 5 and 6) of an external speaker-based apparatus for the generation of oscillatory flow with frequencies in the range of 10 to 1000 Hz in microfluidic devices. Particle tracking of suspended tracer particles is required to determine the fidelity of the harmonic motion as well as for calibrating the range of oscillation amplitudes achievable over the range of operating frequencies. The amplitude-frequency curve for a given volume setting depends primarily on the characteristics of the speaker, which cannot be changed (see discussion of speaker characteristics in representative results for Figure 3A,B). However, for a particular channel design, the oscillatory amplitude can be modified and tuned by appropriately modifying the tubing properties, the liquid viscosity, or combinations thereof. For example, we show in Figure 4 that a larger tube diameter or longer tube length can increase the magnitude of the oscillatory amplitude for the same volume setting. Increasing viscosity, however, decreases the range of oscillatory amplitudes, providing users with a range of amplitudes, extending from 10 µm to 1 mm.
The significant advantage of this method is its ease of assembly, implementation, and operation. The entire cost of the oscillatory driver is less than $60 and its assembly will only take approximately 2 h once the parts are purchased (see Table of Materials). Unlike alternative methods for generating oscillatory flow in microfluidic devices25, this method imposes virtually no design constraints and ensures minimal lead time to implementation. Despite its simplicity, our method allows the user surprisingly precise control of oscillation amplitudes while maintaining the fidelity of both sinusoidal and non-sinusoidal oscillatory waveforms. The technique also generates harmonic motion over a frequency range of two orders in magnitude. Lastly, this technique can be used together with a steady flow component generated by standard microfluidic flow controllers, such as syringe pumps or pressure generators, to generate a high-frequency pulsatile flow. As previously demonstrated22,28, the oscillatory amplitude and frequency are not affected by the presence of a steady transport flow when the steady flow velocity is small compared to the oscillatory flow velocity. This method is therefore ideal for a research laboratory setting.
A corresponding limitation of the method is that the amplitude cannot be set at the desired value. It must be measured and calibrated to the amplitude for a given microfluidic channel. It is currently not scalable and thus not immediately suitable for industrial applications. Further development of this apparatus would involve the design of a simple diaphragm that can be bonded to and actuated by the speaker to permit larger amplitudes and minimize the dependence on the tubing and microfluidic channel.
Overall, this work provides a low-cost, robust, and customizable approach for generating oscillatory flows in microfluidic channels in a relatively unexplored frequency range. This technique has been shown to be useful for the microrheology of Newtonian26 and non-Newtonian27 liquids, enhanced mixing at the microscale28, and inertial focusing in channels of reduced length22. The approach outlined in this work provides an accessible and adaptable methodology to generate purely oscillatory flows, or pulsatile flows when combined with a steady flow from a syringe pump. As a result, this convenient technique can enable the implementation of oscillatory flows into existing research and industrial at the microscale.
The authors have nothing to disclose.
We would like to acknowledge the support given and facilities provided by the Department of Mechanical Science and Engineering Rapid Prototyping Lab at the University of Illinois to enable this work.
Oscillatory Driver Assembly | |||
Alligator-to-pin wire | Adafruit | 3255 | Small alligator clip to male jumper wire (12) |
Aux cable | Adafruit | 2698 | 3.5 mm Male/Male stereo cable 1 m |
Controller chip | Damgoo | TPA3116 | 50w+50w 2 channel audio amplifier (bluetooth and AUX) |
DC adapter | Adafruit | 798 | 12 V DC 1A regulated switching power adapter |
Micro-pipette tip | VWR Signature | 37001-532 | 200 ul micropipette tip |
Silicone sealant | Loctite | 908570 | Clear silicone waterproof sealant (80 ml) |
Speaker | Drok | 6843996 | 4.5 inch 4 Ohm 40 W speaker |
Speaker mount | 3D printed from 'speakermount.stl' in supplementary files | ||
Speaker-to-tube adapter | 3D printed from 'speaketubeadapter.stl' in supplementary files | ||
Microchannel Manufacture | |||
Biopsy punch | Miltex | 15110 | Biopsy punch with plunger (1 – 4 mm) |
Degasser | |||
Disposable cup | |||
Disposable spoon | |||
Glass Slides | VWR Signature | 16004-430 | 3" x 1" pre clean 1 mm thick |
Mold | Si – SU-8 or 3D printed | ||
Oven | Fischer Scientific | Isotemp | |
PDMS resin and cross-linker | Dow Chemical | 4019862 | Sylgard 184 PDMS resin and crosslinker (500 g) |
Polyethylene tubing | Becton Dickinson Intramedic | 427440 | Polyethylene tubing (PE 60 – PE 200) |
Razor blades | VWR | 55411-050 | Single edge industrial razor blades |
RF plasma generator | Electro-Technic Products | BD – 20 | High frequency generator |
Silicone Mold Release | CRC | 03301 | Food Grade Silicon Mold release (16 oz) |
Observation and Characterization | |||
Camera | Edgertronic | SC2+ | |
Lens | Nikon | Plan Fluor 10x | |
Microscope | Nikon | Ti Eclipse manual stage | |
Needles | Becton Dickinson | 305175 | PrecisionGlide 20G |
Syringe | Becton Dickinson | 1180100555 | Monoject 1 ml |
Syringe pump | Harvard Apparatus | Dual syringe programmable syringe pump | |
Tracer Particles | Spherotech | PP-10-10 | Polystyrene tracer particles 1 um |