Innovative Composite PDMS Micropump with Electromagnetic Drive

In this paper, an innovative composite poly-dimethylsiloxane (PDMS) micropump with an electromagnetic drive for biomedical applications that feature low cost and simple assembly is investigated. The developed PDMS micropump based on a reciprocation principle was driven by electromagnetic force that causes cavitation inside a chamber to make ﬂ uid ﬂ ow. A composite PDMS thin ﬁ lm, an iron-particle-dispersed PDMS (IPDP) thin ﬁ lm, is designed, fabricated and driven by electromagnetic force to actuate the micropump. In this work, there are two categories of micropumps including stacked and inlaid types, and each category has four types of micropumps with different geometrical combinations of IPDP and PDMS thin ﬁ lms. The results show that inlaid-type micropumps all have higher ﬂ ow rates than stacked-type micropumps, and for an inlaid-type-I micropump, the largest IPDP thin ﬁ lm area results in its highest ﬂ ow rate among the categories. The inlaid-type-I micropump has a maximum ﬂ ow rate and a backpressure of 1.623 ml/min and 361.84 Pa, respectively, when applying 30 V between 6 and 7 Hz with a low power consumption of 33 mW. As a result of its higher ﬂ ow rate and new IPDP thin-ﬁ lm design, this kind of full PDMS micropump is appropriate for biomedical applications.


Introduction
For years, micro-electromechanical system (MEMS) technology, i.e., the miniaturization of sensors and actuators, has been rapidly developing owing to its growing commercial applications. (1) The microfluidic system (2) based on MEMS has been widely investigated. Microfluidic systems have many applications, such as microradiators for laptops and biomedical devices. In microfl uidic systems, the micropump is the most important unit for popular fi elds, such as micro-Total Analysis System (3) (micro-TAS or μTAS) and Lab-On-a Chip (LOC). (4) In micropumps, the valve unit is usually studied; however, the inconsistency of the frequency of the valve and the applied frequency may cause damage to the valve. Although the fl ow rate of a valveless micropump may not be the largest among various micropumps, the design of the nozzle/diffuser could simplify the manufacture process and pump structures, and avoid the impairment of valves in the micropump after long-term usage. After the nozzle/diffuser (5) unit was invented, a valveless design without a complicated structure is widely adopted using a simple physical reciprocation principle. Olsson et al. (5) adopted the difference pressure drop between cone-shaped geometries to make the fl uid fl ow. Kim et al. (6) utilized a nozzle/diffuser unit with double chambers that was actuated by piezoelectric disks. Yamahata et al. (7) adopted a magnet embedded in the composite valveless micropump and driven by electromagnetic force. Actuation-type micropumps are the key in fl uidic systems, including nonmechanical and mechanical types. Nonmechanical-type micropumps are actuated by the concentration difference between the anions and cations of the fl uid or capillary attraction, including electroosmotic, (8) electrochemical, (9) electrophoretic, (10) electro-hydrodynamic (EHD) (11) and magneto-hydrodynamic (MHD) (12) actuations. Mechanical-type micropumps have been extensively developed using various actuation methods, such as thermo-pneumatic, (13) thermal-bubble, (14) electromagnetic, (15) piezoelectric (16) and electrostatic (17) actuations. Although nonmechanical-type micropumps consist of a simpler structure, they usually drive with specific fluids and actuated electrodes. Mechanical-type micropumps with actuators could drive any working fl uid with a larger fl ow rate, which is mainly different from nonmechanical-type micropumps. Cavity chambers are necessary for most mechanical-type micropumps that make the reciprocation to cause fl uid fl ow. Most mechanical-type micropumps show a larger pressure and a faster response than nonmechanical-type micropumps. Previously, the development of microfl uidic systems was based on conventional semiconductor materials and techniques originally developed for the integrated circuit industry. Using these techniques for developing microfl uidic systems has resulted in not only high cost but also many limitations on fabrication. On the other hand, the recent poly-dimethylsiloxane (PDMS) (18) micromolding technology has been utilized, thereby overcoming such problems. PDMS is one of the most promising polymers used in biomedical devices, such as catheters, tracheoesophageal voice prostheses, finger joints, percutaneous devices, and dentures. (19) PDMS shows good heat stability, oxidation stability, low surface tension, good ventilation and excellent insulation at room temperature; it also shows softness, elasticity and biocompatibility (19,20) at low temperatures; moreover, a PDMS molding with a simple fabrication process and a low cost is very suitable for disposable microfl uidic systems, especially for biomedical applications. In addition, the PDMS microfl uidic system can be easily integrated with other microfl uidic systems within PDMS replication and is applied for large-scale production. It is very easy to fabricate disposable microfl uidic systems because the fabrication procedure of the planar geometry of PDMS replication is easy and inexpensive.
In this study, the PDMS is adopted to fabricate the disposable valveless micropump by electromagnetic actuation to investigate its performance. A specially designed IPDP (iron-particle-dispersed PDMS) thin fi lm is fabricated such that a full PDMS micropump, suitable for biomedical applications, can be achieved. Various geometric designs of micropumps are implemented in this work, and the fl ow rate and back pressure of those micropumps driven at various applied voltages and switch frequencies are measured. Two categories of electromagnetic micropumps were demonstrated in terms of their performance.

Principle and design of valveless micropump
Recently, the valveless micropump has become very popular and its key component is the nozzle/diffuser unit. Figure 1 reveals the principle of the nozzle/diffuser unit in a pump including the supply and pump modes. The volume fl ow rate Q 1 is larger than Q 2 when the chamber is expanded by the actuation of electromagnetic force by an electromagnet, as shown in Fig. 1(a). This is because the inlet fl uid speed (larger arrow besides Q 1 ) is higher than the outlet speed (Q 2 ) while the fl ow passes the same cone-shaped channel but in the reverse direction. A similar function of the nozzle/ diffuser system leads to the volume fl ow rate Q 2 being larger than Q 1 when the chamber collapses, as shown in Fig. 1(b). The result shows that the combination of the above two  nlet Outlet modes makes the fl uid fl ow from the inlet to the outlet. According to the stability map of a diffuser as shown in Fig. 2, (21) the diffuser operates in four different regions depending on the diffuser geometry. In the bistable steady stall (between b-b and c-c lines) and jet fl ow (above c-c line) regions, the fl ow performance is poor to extremely poor, as shown in Fig. 2. Under the line a-a, the nostall region, the fl ow is steady viscous without separation at the diffuser walls and a moderate performance is achieved. In the transitory steady stall region between a-a and b-b lines, the fl ow is unsteady. The minimum pressure loss and maximum pressurerecovery coeffi cient Cp occur in this region, and hence, the diffuser geometry is designed accordingly. Therefore, the diffuser geometry was designed to be L/W 1 = 9 and 2θ = 9.8°, in which W 1 , W 2 and L are respectively the upper base diameter, lower base diameter, and attitude of the cone-shaped nozzle/diffuser system. Olsson et al. (22) investigated the fl ow rate and backpressure from slenderness, L/W 1 , and conical angle in 1999. Kim et al. (23) also adopted L/W 1 and 2θ between the transitory and steady stall regions.

Fabrication of electromagnetic valveless PDMS micropumps
In this work, two categories of electromagnetic micropumps are studied: one is a stacked-type micropump, as shown in Fig. 3(a), and the other is an inlaid-type micropump, as shown in Fig. 3(b). Both micropumps have the same lower section consisting of a chamber, inlet, outlet and nozzle/diffuser unit. The upper section of the inlaid-type micropump includes an inlaid thin fi lm, in which an iron-particle-dispersed PDMS (IPDP) thin fi lm is embedded in a pure PDMS thin fi lm. The upper section of the The fabrication process of the two categories of PDMS valveless electromagnetic micropumps is described as follows. For the PDMS microfl uidic structure, a polymethylmethacrylate (PMMA) mold with the desired geometry including the inlet, outlet, nozzle/diffuser microchannel and chamber was fabricated, and then, the PDMS (polydimethylsiloxane, Sylgard 184 silicone from Dow Corning, USA) solution mixed with resin and hardener in weight ratio of 10:1 was poured into the PMMA mold. After evacuating in a vacuum chamber for 60 to 90 min, the PDMS polymer was baked at 70°C for 180 min and subsequently cooled to room temperature. Finally, the PDMS microfl uidic structure replicas were peeled off the PMMA mold. With this structure, the fabrication processes of stacked-type and inlaid-type micropumps are described respectively in Figs. 5 and 6. For the inlaid-type micropump, the IPDP thin fi lm was then fabricated as follows. The PDMS solution was well mixed with resin, iron particles (Nippon Shiyaku Kogyo K. K., Japan and the average size of iron particles is 55 μm), and hardener in a weight ratio of 10:10:1 such that iron particles are uniformly dispersed in PDMS; this PDMS mixture was then evacuated in a vacuum chamber for 60 to 90 min. Afterward, the mixture was baked at 80°C for 60 min, cooled to room temperature, and fi nally cut into circles with four different diameters, as shown in Fig. 4. After baking and hardening, PDMS can seal iron powder as the IPDP thin fi lm hermetically and completely. The PDMS solution mixed with resin and hardener in weight ratio of 10:1 was poured around the IPDP thin fi lm at the same thickness. This thin fi lm was evacuated and baked to fabricate the inlaid thin fi lm, and cut at the same length× width of the microfl uidic structure (length×width, 55 mm×35 mm). The inlaid thin fi lm was attached to the PDMS microfl uidic structure by the PDMS solution to achieve a micropump. The stacked-type micropump has a processing different from that of the inlaid-type micropump. Initially, an IPDP thin fi lm was fabricated accordingly. Then, the IPDP thin fi lm was stacked onto the pure PDMS thin fi lm and attached to what by the PDMS solution to obtain a stacked thin fi lm. Similarly, the stacked thin fi lm was attached to the microfl uidic structure to achieve a micropump. The fabricated stacked- type-I and inlaid-type-I IPDP micropumps are shown in Figs. 7(a) and 7(b), respectively. The chamber that connects to the right outlet and left inlet of the nozzle/diffuser unit is below the black IPDP thin fi lm.

Measurement
To actuate the micropump, the micropump is driven with an electromagnetic force by connecting an electromagnet, a power supply, a combifl ex (a device to control the switch on-off of power supply) and a function generator to control the switch frequency of the power supply for actuating the inlaid or stacked thin fi lm. The measurement setups of the water fl ow rate and back pressure for the micropump are shown in Figs. 8 and 9, respectively. The air gap between the electromagnet and IPDP thin fi lm is always fi xed at 500 μm, which prevents the IPDP thin fi lm contact with the electromagnet, thereby affecting the defl ection of the IPDP thin fi lm in the expand mode. The applied voltages on the electromagnet are 20 and 30 V. The frequency range of the applied voltage is from several hertz up to 20 Hz. The backpressure is defi ned as the height difference between the outlet port and the inlet port. Before actuating, the micropump and water reservoir are placed at the same height to avoid fl ow by a siphon phenomenon, then the micropump could drive fl uid to fl ow. In terms of the results of Shen and Liu, (24) a micropump with the maximum fl ow rate occurs at the resonant frequency of the entire  Fig. 10. The loss weight above electronic balance is the electromagnetic force when the electromagnet is activated.

Flow rate
Flow rates of the four kinds of inlaid-type micropumps at applied voltages of 20 and 30 V are shown in Fig. 11. The maximum fl ow rate of inlaid-type I is 1.623 ml/min when applying 30 V between 6 and 7 Hz. The maximum fl ow rates of inlaid-types II, III and IV are respectively 1.124, 0.947, and 0.817 ml/min between 6 and 7 Hz as well. The fl ow rates of the four inlaid-type micropumps increase when the applied frequency increases in the range from 0.5 to 7 Hz, and all the maximum fl ow rates were observed between 6 and 7 Hz because the resonant frequency of a micropump is close to this range. However, the fl ow rates of the four inlaid-type micropumps decrease when the applied frequency increases in the range from 7 to 20 Hz. The fl ow rates of all the inlaid-type micropumps approach zero when the applied frequency is larger than 20 Hz. The fl ow rate of the micropumps subjected to the applied voltage of 30 V is larger than that subjected to the applied voltage of 20 V, because the former provides a larger power of 33 mW (current of power supply, 1.1 mA) than the latter (22 mW). The inlaidtype-I micropump has the highest fl ow rate among four kinds of inlaid micropumps; the fl ow rates of the other micropumps from high to low are in the order of inlaidtype micropumps II, III, and IV. The IPDP thin fi lm of a micropump is actuated by an electromagnetic force; therefore, the area of the IPDP thin fi lm is the key parameter affecting electromagnetic force. According to Maxwell's pulling force (25) and Lorentz force formula, (26) the electromagnetic force F is given by where F is the electromagnetic force, B is the magnetic flux density of the electromagnetic fi eld (by measurement, B is 650 mT when the applied voltage is 20 V and B is 850 mT when the applied voltage is 30 V), A is the area of the IPDP thin fi lm (the areas for types I, II, III, and IV are respectively 346.2, 86.5, 38.5, and 9.6 mm 2 ), and μ 0 is the permeability of a vacuum. In terms of eq. (1), the electromagnetic force is affected by the area of the IPDP thin fi lm at the same magnetic fl ux density (B) and the permeability of a vacuum (μ 0 ). For example, both the inlaid-type micropumps I and II have the same B and μ 0 values, but not A (air gap between the electromagnet and the IPDP thin fi lm is always fi xed at 500 μm); hence, the electromagnetic force of the  inlaid-type-I micropump is theoretically four times that of the inlaid-type-II micropump. Depending on the Lorentz force, the larger area provided a higher electromagnetic force at the same electromagnetic fi eld. As shown in Table 1, the measured electromagnetic forces for the inlaid-type micropumps I, II, III, and IV are 178.46, 52.19, 11.18 and 2.75 mN, respectively, when the applied voltage is 30 V. Therefore, the inlaid-type-I micropump has the largest fl ow rate among the four inlaid-type micropumps caused by its largest electromagnetic force. Theoretically, according to eq. (1) with other parameters remaining constant, the electromagnetic force (F) is proportional to the IPDP area (A). However, for the inlaidtype micropumps II and III with an area ratio of 2.3 (9/4), their measured electromagnetic force ratio of 4.7 (52.19/11.18) is not the same as the area ratio. This indicates that the structure of the micropump would affect the measured electromagnetic force. It is found that when the area ratio of IPDP/chamber (both have the same thickness) is around 0.12 ([7 2 × /4]/[20 2 × /4]), which is the case for the inlaid-type-III micropump, the prediction of electromagnetic force would deviate from eq. (1). It is suggested that the deviation is due to the local nonuniform defl ection around a smaller IPDP layer above a larger empty chamber when subjected to electromagnetic force. Therefore, in order to achieve a higher electromagnetic force, the area ratio of IPDP/chamber is suggested to be larger than the critical value of 0.12. Similarly, as compared with the electromagnetic force at 30 V between the inlaid-type-I (178.46 mN) and stacked-type-I micropumps (143.72 mN) in Table 1, the difference in the force is caused by the structure variation underneath the IPDP layer. Figure 12 shows the relationship between the fl ow rate and applied frequency of four kinds of stacked-type micropumps. The maximum fl ow rate of the stacked-type-I micropump is 1.41 ml/min when applying 30 V between 6 and 7 Hz, which is lower than that of the inlaid-type-I micropump. The larger area of the IPDP thin fi lm in the stacked-type-I micropump leads to a larger electromagnetic force as well as a larger power, and results in a fl ow rate higher than those of the stacked-type micropumps II, III, and IV. According to authors' previous results (24) and other researchers' works, (23,27,28) the maximum fl ow rate occurred at the resonant frequency of the micropump due to the maximum defl ection. As shown in Table 1, the larger area of the IPDP thin fi lm leads to a larger electromagnetic force that results in a higher fl ow rate due to a larger defl ection.
In the comparison of Figs. 11 and 12, in addition to the difference in electromagnetic force shown in Table 1, the fl ow rate of the stacked-type micropump that is lower than that of the inlaid-type micropump could be attributed to two other factors: stiffness and mass difference. As mentioned in the previous paragraph, the defl ection of the fi lm above the chamber determines the fl ow rate. Under the same electromagnetic force (IPDP area), the fi lm of the stacked type that consists of both IPDP and PDMS layers has a higher stiffness than the fi lm of inlaid type with only one PDMS layer, thus resulting in a lower defl ection as well as a lower fl ow rate. Furthermore, the stacked-type micropump with an extra weight of the pure PDMS thin film underneath the IPDP thin film constrains the vibration behavior, thereby limiting the driving effect. For example, although both the inlaid-type-I and stackedtype-I micropumps have IPDP thin fi lms of the same size, under the same driving electromagnetic force, the stacked-type-I micropump needs to drive both the IPDP and PDMS thin fi lms underneath, which are 0.2 g heavier than that driven by the inlaid-type-I micropump. By Newton's second law, a heavier mass results in a lower acceleration under the same electromagnetic force. The proper integration of this lower acceleration with respect to time would result in a lower vibration speed as well as a lower frequency of both layers above the stacked-type chamber. This could shift its vibration frequency from the resonant frequency. Therefore, the inlaid-type micropump with no additional weight of the pure PDMS thin fi lm has a higher fl ow rate owing to its higher vibration frequency. Figures 13 and 14 show the relationships between the back pressure and applied switch frequency of the four kinds of inlaid-type and stacked-type micropumps, respectively. The trends of the backpressure of all micropumps in both Figs. 13 and 14 are similar to those of fl ow rates shown in Figs. 10 and 11, respectively, i.e., the inlaidtype-I micropump has the highest back pressure and the stacked-type-IV micropump has the lowest back pressure.

Back pressure
For the inlaid-type-I micropump, its back pressure increases with the increase in applied voltage because a higher voltage provides a stronger electromagnetic force. The back pressure also increases with increasing size of the IPDP thin fi lm for similar reasons; hence, the back pressure of the inlaid-type-I micropump is larger than that of the inlaid-type-II micropump. Both the back pressure and flow rate of the inlaid-type micropump are larger than those of the stacked-type micropump. Of all the micropumps, the inlaid-type-I micropump has the highest fl ow rate and largest back pressure because it has the largest IPDP thin fi lm and the lowest driving inertia that provides the highest driving force for the micropump. Table 2 shows a comparison of the key characteristics of several electromagnetic and piezoelectric micropumps. The former is made of metal and plastic materials, whereas the latter is made of polymers and plastic materials. EsoxPump TM V01 (29) is a polymeric implantable pump that has been adapted from the clinical market. Furthermore, Bartels mP5 (30) and ThinXXS MDP1304 (31) are positive displacement (the thin fi lm above the chamber moves from its stationary state to expand similarly in Fig. 13. Relationships between back pressure and frequency for all inlaid-type micropumps.   Inlaid type  I-30V  I-20V  II-30V  II-20V  III-30V  III-20V  IV-30V  IV- (32) developed a piezoelectric micropump using a deformable elastomeric membrane to drive the micropump. In addition, Pan et al. (33) fabricated two positive-displacement electromagnetic micropumps by PDMS and PCB. Grosjean and Tai (34) at CalTech have fabricated various micropumps using silicon and polymer (Parylene and PDMS). The inlaid-type-I micropump in this study has a larger fl ow rate than Pan's microcoil-driven pump but a lower back pressure and a lower power consumption. A low back pressure resulting in a higher fl ow rate is the advantage of our micropump, in that a lower pressure can actuate a higher fl ow rate.

Pumping effi ciency
Several micropumps with various dimensions and fl ow rates are listed in Table 2 for comparison. There are two kinds of actuation in the table: piezoelectric and electromagnetic. The difference between piezoelectric and electromagnetic actuations is the applied frequency range. In general, the frequency range for the piezoelectric type is from several Hz to a couple of kHz and the frequency range for the electromagnetic type is from 0 Hz to several hundred Hz. However, the electromagnetic driver can work in a noncontact manner, which may be useful for biomedical applications. Our micropump ranks third among the eight micropumps in terms of flow rate. For MEMS applications, the dimensions of the micropump would play an important role. Therefore, we defi ne a new factor named pumping effi ciency by dividing fl ow rate by the dimensions of the entire micropump (effi ciency = fl ow rate/dimensions). A higher effi ciency indicates that a micropump can provide a higher fl ow rate by unit volume. For further nanoapplications, this would be an important performance index. Judging from the pumping effi ciency data in Table 2, the Bartels mP5 micropump has the highest pumping effi ciency, although it does not have the highest fl ow rate. The micropump in this work ranks third among the four electromagnetic pumps in terms of effi ciency. The low effi ciency of our micropump is probably due to the absorption of driving defl ection caused by the low stiffness of the full PDMS micropump. The advantage of our micropump is that iron particles are embedded in PDMS such that the entire micropump is highly biocompatible and can be easily mounted in the human body.

Conclusion
Novel composite PDMS valveless micropumps were fabricated and actuated by electromagnetic force through the bonding of microfluidic structures with stacked or inlaid PDMS thin fi lms composed of IPDP and pure PDMS thin fi lms; and their performance characteristics were characterized. Micropumps that can be driven by adjustable switch frequency, low applied voltage and low power consumption were successfully demonstrated in this study. The fl ow rates and back pressures of both stacked-type and inlaid-type micropumps increase with the voltage applied to an electromagnet in a certain frequency range. This electromagnetic actuated micropump is developed with the advantages of high fl ow rate, low applied voltage, low current and low power consumption. The electromagnetic force and stiffness and mass of the driven layer are three main factors that affect its defl ection to affect fl ow rate and back pressure. With a larger electromagnetic force, a smaller stiffness and a lower mass, inlaid-type micropumps have both a higher fl ow rate and a higher back pressure than stacked-type micropumps. The inlaid-type-I micropump has the highest fl ow rate of 1.623 ml/min and the largest back pressure of 361.84 Pa among the eight kinds of micropumps. Although the inlaid-type-I micropump shows an intermediate pumping effi ciency of 8.43×10 −2 min −1 among the existing micropumps, considering the volume of the micropump with embedded iron particles in the composite fi lm and a noncontact driving mechanism, it is applicable to biomedical devices.