See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/222106333
Valve-less diffuser micropump for microfluidic analytical systems ARTICLE in SENSORS AND ACTUATORS B CHEMICAL · FEBRUARY 2001 Impact Factor: 4.1 · DOI: 10.1016/S0925-4005(00)00644-4
CITATIONS
READS
143
186
5 AUTHORS, INCLUDING: Wouter Van der Wijngaart
Peter Nilsson
KTH Royal Institute of Technology
KTH Royal Institute of Technology
112 PUBLICATIONS 1,392 CITATIONS
156 PUBLICATIONS 5,908 CITATIONS
SEE PROFILE
SEE PROFILE
Peter Enoksson
Göran Stemme
Chalmers University of Technology
KTH Royal Institute of Technology
157 PUBLICATIONS 2,094 CITATIONS
341 PUBLICATIONS 6,559 CITATIONS
SEE PROFILE
All in-text references underlined in blue are linked to publications on ResearchGate, letting you access and read them immediately.
SEE PROFILE
Available from: Peter Enoksson Retrieved on: 15 January 2016
Sensors and Actuators B 72 (2001) 259±265
A valve-less diffuser micropump for micro¯uidic analytical systems Helene Anderssona,*, Wouter van der Wijngaarta, Peter Nilssonb, Peter Enokssona, GoÈran Stemmea a
Department of Signals, Sensors and Systems, Royal Institute of Technology, Instrumentation Laboratory, Osquldas vag 10, S-10044 Stockholm, Sweden b Department of Biotechnology, Royal Institute of Technology, Instrumentation Laboratory, Osquldas vag 10, S-10044 Stockholm, Sweden Accepted 6 September 2000
Abstract The suitability of valve-less micropumps in biochemistry has been shown. Fluids encountered in various biochemical methods that are problematic for other micropumps have been pumped with good performance. The pump is fabricated as a silicon-glass stack with a new process involving three subsequent deep reactive ion etching steps. Some of the main advantages of the valve-less diffuser pump are the absence of moving parts (excluding the pump diaphragm), the uncomplicated planar design, and high pump performance in terms of pressure head and ¯ow rate. In addition, the micropump is self-priming and insensitive to particles and bubbles present in the pumped media. The results show that the valve-less micropump successfully pumps ¯uids within the viscosity range of 0.001±0.9 N s/m2. The micropump is not sensitive to the density, ionic strength, or pH of the pumped media. Effective pumping of solutions containing beads of different sizes was also demonstrated. Living cells were pumped without inducing cell damage and no cell adhesion within the pump chamber was found. No valve-less micropump has previously been reported to pump such a wide variety of ¯uids. # 2001 Elsevier Science B.V. All rights reserved. Keywords: Valve-less micropump; Highly viscous liquids; Bead handling; Living cells
1. Introduction Micropumps are essential components in micro¯uidic analysis systems. A miniaturized laboratory (lab-on-a-chip) must be able to handle ¯uids with a wide variety of properties, i.e. viscosity, density, ion strength, pH, temperature, and surfactants. The micropump's sensitivity to ¯uid properties is dependent on the principle used for ¯uid movement. Today, many presented microchemical analysis systems are based on electroosmotic and electrohydrodynamic pumping [1,2]. The underlying pump principles build on the speci®c ¯uid properties of the pumped medium [3,4]. Electrohydrodynamic pumps use the kinetic energy of ions present in the liquid to create the pump action and electroosmotic pumps use the presence of immobilized surface charges, mobile charges in the sample and an externally applied electrical ®eld. Thus, these pumps are thus inherently dependent on the properties (pH, ionic strength) of the pumped medium, making them unsuited for a large number of biochemical and biological liquids [5]. Therefore, this * Corresponding author. Tel.: 46-8-790-92-36; fax: 46-8-10-08-58. E-mail address:
[email protected] (H. Andersson).
paper focuses on pump principles using diaphragm displacement for generating the required pressure/¯ow. The micropumps having best performance in terms of pressure/¯ow characteristics are passive check valve pumps [6] and valve-less diffuser pumps [7,8]. However, passive check valves induce dif®culties such as clogging and sedimentation and have a rather complicated design [6]. Therefore, they are not suitable for extreme miniaturization. The ®rst valve-less diffuser micropump in silicon was presented in 1997 [8]. It uses diffusers as ¯ow directing elements. Wear and fatigue in the valves are eliminated since the diffuser elements have no moving parts. The risk of clogging is also reduced. The valve-less diffuser pump consists of two diffuser elements connected to a pump chamber with an oscillating diaphragm. The key components of the micropump are the ¯ow directing diffuser elements. One diffuser element is directed from the inlet to the pump chamber and the other diffuser element from the pump chamber to the outlet as illustrated in Fig. 1. The oscillating diaphragm forces the ¯uid through the two diffuser elements. The result is a net transport of ¯uid from the inlet to the outlet due to the difference in the ¯ow resistances in the forward (diffuser) and reverse (nozzle)
0925-4005/01/$ ± see front matter # 2001 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 5 - 4 0 0 5 ( 0 0 ) 0 0 6 4 4 - 4
260
H. Andersson et al. / Sensors and Actuators B 72 (2001) 259±265
Fig. 1. Working principle of the valve-less diffuser micropump (single chamber).
directions (Fig. 1). Some of the main advantages of the diffuser pump are the absence of moving parts (excluding the pump diaphragm), the uncomplicated planar design, high pump performance, in terms of pressure head and ¯ow rate, and the ability to pump a wide variety of ¯uids. Pump reliability is important in achieving continuous pumping and is a challenge for micropumps. Pump reliability involves a number of speci®c features. First, the pump must be able to create an under-pressure, sucking liquid towards the pump inlet. This is called self-priming and can be achieved if the pump has both gas and liquid pumping capabilities. Second, pump ®lling must proceed without trapping any gas pockets, which can be ensured by a proper geometrical design. Reliability also involves successful long term pumping periods, during which small particles or gas bubbles present in the pumped liquid should not decrease device performance. Particle sensitivity is an inherent problem for check valve pumps, in which the moving valves tend to clog [6]. Diffuser pumps do not suffer from these restrictions as they have no moving parts and/or small ¯ow channel geometries. However, diffuser pumps presented earlier [8] are sensitive to gas bubbles and cavitation (i.e. vapor formation resulting from a pressure drop below the vapor pressure). In addition, an extensive priming procedure is required, making them unreliable for practical applications. 2. Design A new valve-less diffuser micropump for pumping of both gas and liquid was designed, fabricated, and evaluated [9]. The pump has a single pump chamber and piezoelectric actuation. It consists of a silicon-glass stack and is fabricated with a new process involving three subsequent deep reactive ion etching (DRIE) steps. Fig. 2 illustrates the novel design features including deep diffusers, a shallow pump chamber, and a thin pump diaphragm. The deep diffusers enable a good pressure/¯ow performance while the combination of a shallow pump chamber and a thin diaphragm result in a high compression
Fig. 2. Schematic of a complete pump in top (A) and side (B) view. The diffuser neck is square and measures 30 mm deep and wide. The shallow chamber depth varies between 5 and 30 mm. The pump chamber diameter measures 6 mm in all designs.
ratio. This allows a number of new features, like handling of both gas and liquid, self-priming and cavitation, and gas bubble tolerance. The thin diaphragm also allows a wide driving frequency range and enables bi-directional (i.e. both forward and reverse) pumping of both liquid and gas. In this study the new valve-less diffuser micropump is evaluated for biochemical applications. Extreme concentrations of ¯uids relevant in a wide variety of procedures within the diverse ®eld of biochemistry (see Table 1) were pumped. Samples containing microbeads and living cells were as well investigated. 3. Microfabrication The pumps were fabricated using a new fabrication sequence involving three subsequent DRIE steps on the frontside of the silicon wafer (see Fig. 3). Prior to etching, three masks are patterned onto the silicon: an oxide mask (a), aluminium (b), and a photoresist mask (c). In the ®rst etch, ¯uid connections are created at the in- and outlet (d). The resist mask is removed and in the second etch the deep diffusers are formed (e). The shallow pump chamber is created during the third etch after removal of the aluminium mask (f). After the oxide is removed, the silicon wafer is anodically bonded to a Pyrex wafer (g). In the last DRIE step, the pump diaphragm is thinned down (using a timed etch stop) and in- and outlet ¯uid connectors are formed on the back side of the silicon (h). Typical pump dimensions are given in Fig. 2. The diffuser neck is square and measures 30 mm deep and wide. The shallow chamber depth varies between 5 and 30 mm. The pump chamber diameter measures 6 mm in all designs. After dicing the wafers, a 4 mm diameter, 0.1 mm thick piezo disc is glued and contacted on the pump diaphragm. External polyethylene (PE) tubes are
H. Andersson et al. / Sensors and Actuators B 72 (2001) 259±265
261
Table 1 Fluid properties encountered in molecular biology Property Viscosity Density Ion strength pH Temperature Surfactants
Range
Application source ÿ3
1 10 ±11 10 1±2 g/ml 0±1 M 1±6 6±13 ÿ20±948C 0.05±1.0%
ÿ3
2
N s/m
used as ¯uid connectors and are ®xed onto the chip in a multi-step procedure as shown in Fig. 4. A mask-de®ned silicon surface roughening is performed around the ¯uid openings by DRIE to ensure good adhesion of the PE tubes.
Glycerol (50%, 11 10ÿ3 N s/m2) for handling of enzyme stock solutions Glycerol (50% 1.25 g/ml) for handling of enzyme stock solutions Monovalent salt for hybridizations Acetic acid for neutralization NaOH for denaturation Long term storage and DNA denature Tween for minimization of surface adhesion
A guide wire is used to align the PE tube with the ¯uid openings on the chip. The silicon-glass stack is shortly heated to generate local melting of the PE tube onto the chip. To give additional strength to the assembly, the interface between the chip and the PE tubes is covered with epoxy glue. 4. Experimental The pump chip was mounted under a standard light microscope. Low concentrations of microbeads (2.7 mm in diameter) were added to the ¯uid before pumping to enable the ¯ow pattern inside the pump chamber to be studied. The driving voltage of the piezoelectrical disc was kept below 100 V (electrical breakthrough limit of the piezodisc). The frequency was adjusted to give maximal ¯ow rate for the different liquids. The ¯ow was determined at the in- and outlet at atmospheric pressure by measuring the speed of the liquid column.
Fig. 3. Schematic of the fabrication process.
Fig. 4. Attachment of the fluid connectors by using a new melt-on method.
262
H. Andersson et al. / Sensors and Actuators B 72 (2001) 259±265
The pump ¯ows reported here are much lower than the values presented earlier [8]. However, this is what can be expected considering the small size of the diffuser neck. Future versions will include more optimal diffuser designs for enhanced ¯ow performance. However, the desired ¯ow rate seldom exceeds 1 ml/min for m-TAS. The ¯ow pattern inside the pump chamber was visualized by low concentrations of microbeads in the liquids. Vortexlike ¯ow patterns occur in the pump under all conditions. The number of vortices and the direction in which they turn is dependent on driving frequency and diaphragm amplitude. Generally, the higher frequencies and/or diaphragm amplitudes result in a higher number of vortices and higher ¯ow velocities. A number of different ¯ow patterns observed inside the pump chamber are shown in Fig. 6. The observed ¯ow patterns in the pump chamber indicates that the micropump could function as a micromixer if two inlets are fabricated [11]. The ¯ow patterns described above have a strong in¯uence on the gas bubble behavior inside the pump chamber. Bubbles present in the pumped medium or formed in the pump chamber are washed out of the pump chamber by the vortices. Hence, bubbles do not affect the pump action signi®cantly. High concentrations of microbeads (6:8 108 beads/ml) of two different sizes, 2.7 and 5.5 mm in diameter, have been pumped. The results are presented in Table 3 and show that the pump performance is insensitive to particles (? < 5:50 mm) in the pumped media. Beads are routinely used as the mobile solid phase in medical diagnostics, microbiology, cancer research, immunology and molecular biology for separation, synthesis and detection of molecules. Handling cells on a microchip is important for the sample preparation steps involving, for example, cell sorting and cell lysis. Insect cells and human myeloid leukemia cells were successfully pumped with the new valve-less diffuser micropump. The results are presented in Table 4. The cell suspension contained approximately 3 106 cells/ml. Both cell lines grow in suspension and have cell sizes of about 10± 15 mm. Aliquots of the pumped cell solution were collected at the outlet (every 10 min during 1 h) and the cell viability
4.1. Chemicals and cell lines The pump characteristics for the following samples were investigated: acetic acid 16.7 M, sodium chloride 5 M, glycerol (100%), mineral oil (100%), Tween 20 (100%), paramagnetic beads 2.7 mm in diameter (Dynal, Norway), polystyrene beads 5.5 mm in diameter (Bangs Laboratory, US), insect cells (Sf9, American Type Culture Collection, VA, US), and human myeloid leukemia cells (K-562, American Type Culture Collection, VA, US). 5. Results and discussion The suitability of the valve-less micropump in biochemistry has been shown. Fluids encountered in various biochemical methods (Table 1), that are problematic for other micropumps, have been pumped with good performance. The results are presented in Table 2. Bi-directional pumping was achieved for all the liquids by altering the frequency and amplitude. Hence, the ¯ow directing capability of the diffusers are frequency and amplitude dependent. A deeper understanding of how the diffuser elements function at different driving frequencies and amplitudes are still missing. No clear correlation between the pump rate and the density of the pumped ¯uids could be found. In Fig. 5, the maximum pump rate versus the viscosity of the pumped ¯uids is plotted. It is obvious that ¯uids with higher viscosity have lower pump rates. The pump performance of the valve-less micropump does not depend on the pH of the pumped media at all. For example, concentrated acetic acid has been successfully pumped (Table 2) and the other ¯uids pH used here were in the pH range of 4±8. The valve-less micropump can handle samples of both high (5 M NaCl) and low (DD water) ionic strength without problems (see Table 2). Electroosmotic pumps, for example, are dependent on the ionic strength of the pumped medium. If a network of intersecting channels is used to pump samples of different ionic strength, the electroosmotic ¯ow will become unbalanced and cause problems in the system [10].
Table 2 An overview of the pump resultsa Symbol in Fig. 5
* & ~ ^ &
Pumped fluid
Glycerol Mineral oil Tween 20 Acetic acid NaCl DD water
r (g/ml)
1.25 0.88 1.095 1.05 1.00 0.998
m (N s/m2)
0.9 0.5 0.4 1.22 10ÿ3 ± 1.00 10ÿ3
Amplitude (V)
97 97 97 97 97 97
Forward pumping
Reverse pumping
f (Hz)
Pump rate (ml/min)
f (Hz)
Pump rate (ml/min)
400 400 400 700 700 700
0.3 0.4 0.8 2.1 2.33 2.30
1000 1000 1000 1500 2000 2000
0.3 0.4 0.9 1.4 1.5 1.6
H. Andersson et al. / Sensors and Actuators B 72 (2001) 259±265
263
Fig. 5. The maximum pump rate vs. the viscosity is plotted, ( ) glycerol; (*) mineral oil; (&) Tween 20; (~) acetic acid; (&) DD water.
was veri®ed by methylene blue staining. No cell rupture was detected. The cells did not adhere to the internal surfaces of the pump chamber due to high ¯ow velocities inside the pump chamber. Cells and particles have previously been manipulated in micro¯uidic devices by pressure, dielectrophoresis, and electroosmosis [12±14]. However, these techniques may suffer from problems with sample dilution or electrophoretic damage to sensitive cells [13]. By using the valve-less diffuser micropump high voltages, high frequencies, strong shear forces and high pump temperatures are avoided. In addition, there is no need for sample dilution since the pump can handle highly particle loaded ¯uids.
Transport of highly viscous and particle loaded ¯uids is challenging due to adhesion forces and sedimentation [15]. Dead volumes, i.e. low ¯ow velocity regions, are preferred regions of adhesion and sedimentation and must therefore, be avoided. The valve-less micropump presented here has no dead volume and can handle highly viscous ¯uids well. In addition, the vortex-like ¯ow patterns reduce the risk of sedimentation. The temperature in the pump chamber is important when pumping living cells or sensitive bioliquids. The pump chip temperature was measured during pumping (of water) at two different frequencies, 1 and 3 kHz, with an actuation voltage of 90 V for 30 min. The temperature increased from 1.1 (5%) to 21.98C at 1 kHz and 2.4 (11%) to 24.18C at 3 kHz. These temperatures are well below the critical temperature for most biological samples. 6. Conclusions
Fig. 6. Schematic drawing of some of the commonly observed flow patterns. The black arrows symbolize the diffusers. The grey arrows the observed flow pattern. Some places inside the pump chamber with low flow velocity are indicated.
Today, there is no micropump available which can handle wide variety of ¯uids and can be easily integrated on chip. Therefore, it is very common to use external pumps for m-TAS. The micropump presented here can pump almost any sample relevant in biochemistry and biology, including samples containing beads and living cells. The micropump also makes use of an uncomplicated microfabrication process allowing it to be embedded into m-TAS chips.
264
H. Andersson et al. / Sensors and Actuators B 72 (2001) 259±265
Table 3 Pump results for microbeadsa Pumped bead solution
Magnetic beads (Dynal, Norway) Polystyren beads (Bangs Lab., US) a
1 (mm)
2.8 5.50
C (beads/ml)
Amplitude (V)
6.78 108 1.05 108
97 97
Forward pumping
Reverse pumping
f (Hz)
Pump rate (ml/min)
f (Hz)
Pump rate (ml/min)
700 700
2.3 2.28
1800 1700
1.8 1.9
Bead diameter (1), concentration (C), and pump frequency (f).
Table 4 Pump results for cellsa Pumped cell suspension
Insect cells (Sf9, ATCC, US) Human myeloid leukemia cells (K-562 ATCC, US) a
1 (mm)
10±15 8±10
C (cells/ml) Amplitude (V)
3 106 3 106
90 90
Forward pumping
Reverse pumping
f (Hz)
Pump rate (ml/min)
f (Hz)
Pump rate (ml/min)
300 300
1.9 2.0
3000 3000
2.2 2.0
Cell diameter (1), concentration (C), and pump frequency (f).
Acknowledgements The authors would like to thank Mikael Henriksson at the Department of Medical Biochemistry and Biophysics, Karolinska Institutet, Stockholm, Sweden and Eva Bertram at the Royal Institute of Technology, Department of Biotechnology, Stockholm, Sweden for kindly providing the K-562 and Sf9 cells, respectively.
[11]
[12] [13]
References [1] A. Manz, C. Effenhauser, N. Burggraf, J. Harrison, K. Seiler, K. Fluri, Electroosmotic pumping and electrophoretic separations for miniaturized chemical analysis systems, J. Micromech. Microeng. 4 (1994) 257±265. [2] S. Bart, L. Tavrow, M. Mehregany, J. Lang, Microfabricated electrohydrodynamic pumps, Sens. Actuators A21-A23 (1990) 193±197. [3] S. Shoji, M. Esashi, Microflow devices and systems, J. Micromech. Microeng. 4 (1994) 157±171. [4] A. van der Berg, T. Lammerink, Micrototal analysis systems: microfluidic aspects, integration concept and applications, Topics Curr. Chem. 194 (1998) 21±49. [5] C. Mastrangelo, M. Burns, D. Burke, Microfabricated devices for genetic diagonstics, Proc. IEEE 86 (1998) 1769±1786. [6] P. Gravesen, J. Branebjerg, O. Sondergard Jensen, Microfluidics: a review, J. Micromech. Microeng. 3 (1993) 168±182. [7] E. Stemme, G. Stemme, A valve-less diffuser/nozzle based fluid pump, Sens. Actuators A39 (1993) 159±167. [8] A. Olsson, P. Enoksson, G. Stemme, E. Stemme, Micromachined flatwalled valve-less diffuser pumps, J. Microelectromech. Syst. 6 (1997) 161±166. [9] W. van der Wijngaart, H. Andersson, P. Enoksson, G. Stemme, The first self-priming and bi-directional valve-less diffuser micropump for both liquid and gas, in: Proceedings of the The Thirteenth Annual International Conference on Microelectromechanical Systems, MEMS 2000, Miyazaki, Japan, 23±27 January, 2000, pp. 674±679. [10] X. Qiu, L. Hu, J. Masliyah, J. Harrison, Understanding fluid mechanics within electrokinetically pumped microfluidic chips, in:
[14] [15]
Proceedings of the 1997 International Conference on Solid State Sensors and Actuators, Digest of Technical Papers, Chicago, 16±19 June, 1997, pp. 923±926. Z. Yang, H. Goto, M. Matsumoto, T. Yada, Micromixer incorporated with piezoelectrically driven valve-less micropump, in: Proceedings of the m-TAS '98 Workshop, Banff, CAN, 13±16 October, 1998, pp. 177±180. P. Wilding, L. Kricka, J. Cheng, G. Hvichia, M. Shoffner, P. Fortina, Integrated cell isolation and polymerase chain reaction analysis using silicon microfilter chambers, Anal. Biochem. 257 (1998)95±100. S. Fiedler, S. Shirly, T. Schnelle, G. Fuhr, Dielectrophoretic sorting of particles and cells in a microsystem, Anal. Chem. 70 (1998) 1909± 1915. P. Li, J. Harrison, Transport, manipulation, and reaction of biological cells on-chip using electrokinetic effects, Anal. Chem. 69 (1997) 1564±1568. N. Schwesinger, S. Bechtel, Micropump for viscous liquids and muds, in: Proceedings of the SPIE Conference on Microfluidic Devices and Systems, Vol. 3515, 1998, pp. 40±45.
Biographies Helene Andersson was born in 1974 in Hudiksvall, Sweden. She received her MSc degree in Molecular Biotechnology in 1998 from Uppsala University, Sweden. In the beginning of 1999 she started her PhD studies at the Department of Signals, Sensors and Systems at the Royal Institute of Technology, Stockholm, Sweden. Her main research areas are microfluidics, micrototal analysis systems, micropumps, and nanochemistry. Wouter van der Wijngaart was born in 1973 in Lokeren, Belgium. He received his MSc in Elctrotechnical Engineering in 1996 at the University of Leuven, Belgium. He is currently working on his PhD studies at the Department of Signals, Sensors and Systems at the Royal Institute of Technology, Stockholm, Sweden. His main research areas are microfluidics and microfluidic actuators. Peter Nilsson received his PhD degree in 1998 in Biotechnology from the Royal Institute of Technology in Stockholm, Sweden. Currently he is working as a researcher at the Department of Biotechnology, Royal Institute of Technology. His main research interest is in technology
H. Andersson et al. / Sensors and Actuators B 72 (2001) 259±265 development and utilization of DNA microarrays for gene expression analysis and sequence based DNA analysis. He also has interest in biosensortechnology applied to hybridization analysis and mutation detection. Peter Enoksson was born in Lindesberg, Sweden, on April 19, 1957. He received the MSc degree in Engineering Physics in 1986, the Licentiate of Engineering in 1995 and the PhD in 1997 all from the Royal Institute of Technology, Stockholm, Sweden. In 1997 he was appointed Assistant Professor at the silicon sensor research group at the Department of Signals, Sensors and Systems at the Royal Institute of Technology. His research is in the field of resonant silicon sensors and actuators, especially for fluid applications.
265
GoÈran Stemme was born in Stockholm, Sweden, on February 4, 1958. He received the MSc degree in electrical engineering in 1981 and the PhD degree in solid state electronics in 1987, both from the Chalmers University of Technology, Gothenburg, Sweden. In 1981, he joined the Department of Solid State Electronics, Chalmers University of Technology, Gothenburg, Sweden. There, in 1990, he became an Associate Professor (Docent) heading the silicon sensor research group. In 1991, Dr. Stemme was appointed Professor at The Royal Institute of Technology, Stockholm, Sweden. He heads the Instrumentation Laboratory at the Department of Signals, Sensors and Systems. His research is devoted to sensors and actuators based on micromachining of silicon. Dr. Stemme is a subject editor for the IEEE/ASME Journal of Microelectromechnical Systems.
