ACES Characterization of Surface Micromachined Microfluidic Devices

© International Microelectronics And Packaging Society

The International Journal of Microcircuits and Electronic Packaging, Volume 24, Number 1, First Quarter, 2001 (ISSN 1063-1674)

Intl. Journal of Microcircuits and Electronic Packaging

30

ACES Characterization of SurfaceMicromachined Microfluidic DevicesRyszard J. Pryputniewicz, Paul Galambos*, Gordon C. Brown, Cosme Furlong, and Emily J.Pryputniewicz

NEST – NanoEngineering, Science, and TechnologyCHSLT - Center for Holographic Studies and Laser micro-mechaTronicsMechanical Engineering DepartmentWorcester Polytechnic InstituteWorcester, Massachusetts 01609-2280Phone: 508-831-5536Fax: 508-831-5713e-mail: [email protected]

* Intelligent Micromachines DepartmentSandia National LaboratoriesAlbuquerque, New Mexico

Abstract

Recent breakthroughs in surface micromachining led to development of the first generation surface micromachined microfluidicdevices. The simplest microfluidic device is a microfluidic channel (microchannel). A typical surface micromachined microchannelis about 2 µm deep, its width and depth depend on the specific application, while its top is less than 1 µm thick silicon nitridemembrane. Almost all microfluidic devices are concerned with flow through very small passages and unintentional restrictions thatlead to high-pressure gradients within the microchannels, which, in turn, cause deformations of the membranes. Knowledge of thesedeformations facilitates design and optimization of the microfluidic devices. The researchers have developed a hybrid approach foraccurate and precise characterization of microfluidic devices. This approach is based on Analytical, Computational, and Experi-mental Solutions (ACES) methodology. The experimental aspects of this methodology are based on laser interferometric measure-ments yielding displacement and deformation fields, while analytical and computational aspects are based on exact (closed form) andapproximate (FEM) solutions, respectively. Comparison of analytical, computational, and experimental results indicates correlationto within 1%, for maximum deformations on the order of 1 µm. Continued advances in the methodology for characterization of themicrofluidic devices will lead to development of robust devices, which will allow their integration with electronic actuation to enablea broad range of revolutionary new applications achievable with surface micromachining.

Key words:

MEMS, Microfluidics, Surface Micromachining, Microchannel,ACES Methodology, Laser Interferometry, and OptoelectronicLaser Interferometry Microscope.

1. Introduction

The field of microfluidics is currently undergoing rapid growthin terms of new device and system developments. As a result, avariety of microfluidic devices are being developed for applica-tions ranging from Micro-Total (chemical) Analysis System(µTAS)1 to ink jet printing2. These devices are fabricated using awide range of technologies including bulk micromachining, high

ACES Characterization of Surface Micromachines Microfluidic Devices


© International Microelectronics And Packaging Society 31

aspect ratio (HAR) micromachining, laser micromachining, and,more recently, surface micromachining.

Surface micromachining has significant potential advantagesover other fabrication techniques in some microfluidic applica-tions3. One advantage is volume minimization. A typical bulkmicromachined microfluidic device has a channel depth on theorder of 100 µm, width of 500 µm, and 1000 µm long channel,or a volume on the order of 50 nl. A typical surfacemicromachined channel is only 2 µm deep, 200 µm wide, and1000 µm long, and has a volume of 0.4 nl, which is more than 2orders of magnitude smaller than that of a typical bulkmicromachined channel. In applications where volume minimi-zation is important (such as µTAS), this difference may be sig-nificant in reducing mixing times and reagent requirements.Furthermore, surface micromachined microfluidic devices canbe potentially integrated with electronics to produce electrofluidicdevices. In addition, integration of microfluidics onto a singlesilicon substrate containing MEMS can lead to a new class ofdevices, electro-microfluidic MEMS on a chip3. This class ofdevices has a potential to accomplish a wide variety of functions,such as, power generation, µTAS, hydraulic actuation and con-trol, in a very compact package.

However, in order to use these devices as part of a larger sys-tem on a chip or to interface them with devices external to thechip, performance characteristics of microfluidic device must bequantified. The simplest microfluidic device is a microfluidicchannel (microchannel). This paper addresses design, fabrica-tion, and characterization of a microchannel.

2. Design and Fabrication

Although surface micromachined microfluidic devices offersignificant advantages over devices produced by other method-ologies, the very small sizes of channels pose difficulties in de-vice and system packaging and characterization. In order to ex-plore the potential advantages, Sandia has developed first gen-eration surface micromachined microfluidic devices, Figure 1,containing a number of microchannels.

Figure 1. Sandia microfluidic test-package containingmultiple devices.

A typical surface micromachined microchannel, Figure 2, isabout 2 µm deep, its bottom and sides are lined up with 0.3 µmthick silicon nitride while its top is less than 1 µm thick siliconnitride membrane.

Figure 2. Single microchannel with inlet and outlet vias.

At the beginning of fabrication of microfluidic channels, ther-mal oxide (TEOS) hard mask is patterned onto a silicon waferand 2 µm trenches are etched in silicon4. After etching thetrenches, the TEOS mask is stripped and 0.3 µm thick layer oflow stress silicon nitride is deposited using Low Pressure ChemicalVapor Deposition (LPCVD). Then, a sacrificial oxide (SACOX)refill is deposited and trenches are chemical-mechanically pol-ished (CMP) flat. Following CMP, 0.8 µm thick low stress sili-con nitride membrane (channel cover) is deposited over theSACOX. The membrane is patterned with etch release holesusing another TEOS hard mask, and the release holes are etchedusing a dry etch. After stripping the hard mask, the structuresare released in an acid bath; the solution is highly selective foretching the SACOX while not affecting the silicon nitride. Fi-nally, the etch release holes are sealed using another LPCVDsilicon nitride deposition.

The resulting channels are approximately 2 µm deep. Thesechannels are 200 µm wide at the inlet and outlet and neck downto various widths in between, depending on the specific design.For example, the narrow section of the microchannel shown inFigure 2 is 2 µm deep, 10 µm wide, and 200 µm long.

In order to introduce fluid into the channel, inlet and outletvias, Figure 2, are etched through the wafers from the back usinga Deep Reactive Ion Etch (DRIE) process.

3. Characterization Methodology

Almost all microfluidic devices are concerned with flowthrough very small passages and unintentional restrictions, thatlead to high-pressure gradients within the microchannels, which,in turn, cause deformations of the membranes. Knowledge ofthese deformations facilitates design and optimization ofmicrofluidic devices.

The researchers have developed a hybrid approach for accu-rate and precise characterization of microfluidic devices. Thisapproach is based on Analytical, Computational, and Experi-mental Solutions (ACES) methodology5. In this paper, the ex-perimental aspects of the ACES methodology are based on opto-electronic laser interferometric microscope (OELIM) method,which measures deformations of the microfluidic devices withnanometer accuracy6, while analytical and computational aspectsare based on exact (closed form) and approximate (FEM) solu-tions7, respectively.

200µm




32

( ) ( ) ( )[ ]{ } ,,cos1,2, 0 ii yxyxIyxI θφ ++= (1)

where the subscript i (= 1, 2, ....., n) indicates the i-th interfero-gram – with n being the total number of interferograms, x,y arethe spatial coordinates, ( )yxI ,0 is the average light intensity,

( )yx,φ is the unknown phase, and iθ is the known phase stepdefined as follows,

.2

1πθ −= ii (2)

Based on the set of n equations of the type of Equation (1), spa-tial phase distribution can be determined employing the funda-mental equation,

!!!!!!!"(x, y) = "!(x, y)[I1(x, y), I2(x, y), ...., In (x, y)], (3)

and used to compute displacement vector fields ( )yx,L from thefollowing relation,

( ) ( ) ( ) ( )[ ],,,,,, yxyxyxyx φKLL = (4)

where ( )yx,K is the sensitivity vector defining illumination andobservation (i.e., laser light propagation) geometry of the OELIMsystem used to record interferograms of the microfluidic devices.

3.2. Analytical and Computational MethodsAnalytical methods are characterized by exact, closed form,

solutions and make use of infinitesimal elements in modelingthe microfluidic device. Exactness of the analytical results de-pends on assumptions made during the solution.

Computational methodologies make use of finite size elementsin discretization of the physical domain and provide approxi-mate solutions. The degree of approximation depends on thetype and size of the elements used. In this paper, computationalmodeling was performed using Finite Element method (FEM).

4. Results

Typical interferogram of a deformed membrane of amicrofluidic device, obtained using the OELIM method, is shownin Figure 4. The deformations were caused by loadings pro-duced by compressed air supplied to the microfluidic device.

Quantitative interpretation of interferometric fringe patternshown in Figure 4 yields maximum deformation of 1.048 µm.The analytically determined deformation of 1.044 µm, for thesame loading conditions of the microfluidic device, as those usedto record the interferogram of Figure 4, is in good agreement,well within the uncertainties based on the analytical model - with

3.1. OELIM MethodIn the OELIM method, a beam of collimated coherent light is

brought into the system and is directed into a spatial filter (SF)assembly consisting of a microscope objective and a pinhole fil-ter, Figure 3. The resulting, expanded, light field is then colli-mated by lens L1, and redirected by the directional beam splitter(DBS) through the long distance microscope objective lens (MO)to illuminate the microfluidic device. The proximal beam split-ter (PBS) is placed close to the device. The reflected light istransmitted back through MO, DBS, and the relay lens to theCCD camera.

Figure 3. OLIM system for quantitative measurements ofmicrofluidic devices-optical configuration: SF/LI is thespatial filter/collimating lens assembly “shaping” theillumination beam, DBS is the directional beam splitter, MOis the microscope objective, PBS is the proximal beam splitter,CCD is the host computer controlled image acquisitioncamera.

Illumination of the microfluidic device, by the laser used inthe OELIM system, allows recording of the interferograms of thedevice under its operating condition. These recordings are madeusing phase stepping technique to acquire several interferogramsthat describe a specific state of the object; a known phase step isintroduced for each interferogram8,9. Then, the sequence of thephase-stepped interferograms is processed to obtain detailed quan-titative information on the displacements and deformations ofthe microfluidic devices.

Spatial intensity distribution, ( )yxI , , within the interfero-grams recorded by the CCD camera can be described as follows,

Mirror




Figure 6. FEM determined deformations of the membraneover half of the same microfluidic device as that shown inFigures 4 and 5. Magnitude of the maximum FEMdetermined deformation of the membrane, shown in thisFigure, is 1.059µm.

Representative results shown in Figures 4 to 6 indicate thatgood correlation is obtained between analytical, computational,and experimental results if geometry, dimensions, material prop-erties, initial conditions, boundary conditions, and loading con-ditions are known. However, if any of these parameters changeand accuracy of their values is not known, correlation betweenthe results may not be easy. Nevertheless, differences betweenvarious results can be vividly noticed.

For example, Figure 7 shows a microchannel and its inlet viawith a drop of water trapped at the entrance to the narrow sec-tion. This drop forms an unintentional restriction in themicrochannel and obstructs flow of air from left to right. As aresult of this restriction, high-pressure gradients develop withinthe microchannel and these, in turn, cause deformations of themembrane, subject to the boundary conditions associated withthe presence of the drop of water at the inlet to the narrow sec-tion.

Figure 7. Detail of a microchannel and its inlet via, notedrop of water trapped at the entrance to the 100µm narrowsection.

The microchannel of Figure 7 was loaded by air at 4 psig,OELIM fringe patterns of the deformed membrane were recorded,and their quantitative interpretation yielded results shown in Fig-ure 8. The displacements shown range from 0 nm to 722 nm.

the experimentally measured deformation. FEM results, basedon the model developed during this study, show that the corre-sponding maximum deformation is 1.059 µm, Figure 5.

Figure 4. OELIM interferogram of a deformed membraneof the microfluidic device of Figure 2, showing part of themembrane to the right of the 10µm narrow section. Maximumdeformation of the membrane, determined from thisinterferogram, is 1.048 µm.

Figure 5. FEM determined deformations of the same part ofthe membrane of the microfluidic device as that shown inFigure 4. Maximum FEM determined deformation of themembrane, shown in this Figure, is 1.059µm.

Comparison of the analytical, computational, and experimentalresults, for the case shown in Figures 4 and 5, indicates that thecorrelation between them is within 1%. Similar ACES correla-tion was also obtained by comparing the results obtained for sev-eral other cases of different loading conditions applied to themicrofluidic devices.

Figure 6 shows FEM determined deformations of the entirepart of the membrane to the right of the 10 µm narrow section(i.e., half of the microfluidic device) of the same microfluidicdevice as that displayed in Figures 4 and 5. Clearly, maximumdeformations of 1.059 µm are seen in the center of the mem-brane with the narrow section showing negligible deformations.




34

Figure 8. Representative deformations of the microfluidicmembrane, based on the OELIM fringe pattern correspondingto a load caused by air supplied at 4 psig: 3D planarrepresentation (a) and 3D isometric representation (b).Displacements of the membrane range from 0 to 722 nm andtheir distribution is affected by the drop of water (see Figure7) trapped at the entrance to the narrow section.

The FEM determined deformations of the membrane of themicrofluidic device shown in Fig. 7, excluding the water drop,are shown in Figure 9. For this condition, the maximum defor-mation is 1.045 µm, while the analytical model indicates 1.018µm.

Figure 9. Deformations of the membrane of the microfluidicdevice shown in Figure 7, subjected to air pressure of 4 psig,based on FEM half-model computations: 3D planarrepresentation (a) and 3D isometric representation (b),corresponding to Figures 8a and 8b, respectively. Since thedrop of water (see Figure 7) was not included in the model,magnitude and distribution of the FEM computeddisplacements agree with the analytical results, but differ fromthe experimental ones shown in Figure 8.

Figures 10 and 11 show longitudinal and transverse profilesof the deformed membrane of Figures 7 and 8. These profilesclearly indicate influence of the water drop, trapped at the en-trance to the narrow section of the microchannel.

Figure 10. Longitudinal profile of the deformed membrane,measured along line L shown in Figure 8a.

Figure 11. Transverse profiles of the deformed membrane,measured along lines T1 through T4 shown in Figure 8a:symbols indicate actual data points measured from theOELIM interferogram, while continuous lines displayanalytical representations of the corresponding transverseprofiles.

5. Conclusions

A new hybrid approach for accurate and precise quantitativecharacterization of microfluidic devices was presented in thiswork. This approach is based on ACES methodology. In thisstudy, deformations of surface micromachined microfluidic de-vices, subjected to different loading conditions, were character-ized. ACES correlation between the analytical, computational,and experimental results was within 1%, for an unobstructedchannel. The ACES methodology is also applicable to charac-terization of the microfluidic devices fabricated by other meth-ods than surface micromachining. In addition, it is applicable tocharacterization of microdevices other than microchannels.

In summary, continued advances of methods for characteriza-tion of the microfluidic devices will lead to introduction of non-simple boundary conditions into the analytical and computationalmodels. Following experimental validation of these models, newtools for development of robust microfluidic devices will be avail-able. These tools will allow integration of the microfluidic de-vices with electronic actuation to enable a broad range of revolu-tionary new applications achievable with surface micromachining.




Acknowledgments

The microfluidic devices were fabricated at and provided bySandia National Laboratories. Sandia is a multiprogram labora-tory operated by Sandia Corporation, a Lockheed Martin Com-pany, for the United States Department of Energy under Con-tract DE-AC04-94AL85000.

This study was partially supported by the ME-CHSLT NESTProgram at WPI.

References

1. D. J. Harrison, “Micro-Total Analysis Systems”, Kulver Aca-demic Publishers, Boston, Massachusetts, 1998.

2. S. Kamisuki, “A Low Power Electrostatically Driven Com-mercial Inkjet Head,” MEMS’98, ASME, New York, 1998.

3. P. Galambos, W. P. Eaton, R. Shul, C. G. Willison, J. J.Sniegowski, S. L. Miller, and D. Gutierrez, “SurfaceMicromachined Microfluidics: Design, Fabrication, Packag-ing, and Characterization,” MEMS’99, ASME, New York,pp. 441-448, 1999.

4. E. J. Garcia and J. J. Sniegowski, “Surface MicromachinedMicroengine,” Sensors and Actuators A, Vol. 48, pp. 203-214, 1995.

5. D. R. Pryputniewicz, “ACES Approach to the Developmentof Microcomponents”, M.S. Thesis, Worcester PolytechnicInstitute, Worcester, Massachusetts, 1997.

6. G. C. Brown, “Laser Interferometric Methodologies for Char-acterizing Static and Dynamic Behavior ofMicroElectroMechanical Systems (MEMS)”, Ph.D. Disser-tation, Worcester Polytechnic Institute, Worcester, Massachu-setts, 1999.

7. C. Furlong, “Hybrid, Computational and Experimental, Ap-proach for the Efficient Study and Optimization of Mechani-cal and Electromechanical Components”, Ph.D. Disserta-tion, Worcester Polytechnic Institute, Worcester, Massachu-setts, 1999.

8. G. C. Brown and R. J. Pryputniewicz, “New Test Methodol-ogy for Static and Dynamic Shape Measurements ofMicroelectromechanical Systems”, Optical Engineering, Vol.39, pp. 127-136, 2000.

9. C. Furlong and R. J. Pryputniewicz, “Absolute Shape Mea-surements Using High-Resolution Optoelectronic HolographyMethods,” Optical Engineering, Vol. 39, pp. 216-223, 2000.

About the authors

Ryszard J. Pryputniewicz, educatedboth in Poland and the United States, isProfessor of Mechanical Engineeringand founding Director of the Center forHolographic Studies and Laser micro-mechaTronics (CHSLT) at WorcesterPolytechnic Institute (WPI) in Worces-ter, Massachusetts, since 1978; previ-ously, a faculty member and Director ofthe Laser Research Laboratory at theSchool of Engineering and the Health

Center of the University of Connecticut (6 years); member of theAerospace technical staff (4 years). His research interests con-centrate on theoretical and applied aspects ofMicroElectroMechanical Systems (MEMS), smart sensors andstructures, and, in particular, holographic interferometry. In thiswork, he emphasizes unification of analytical, computational,and experimental solution (ACES) methodologies, especiallywhen they can be merged to provide solutions where none wouldbe obtainable otherwise, to ease the solution procedure, or to at-tain improvements in the results.

Cosme Furlong received his Me-chanical Engineering Degree from theUniversity of the Américas – México,in 1989, and his Master of Science andPh.D. Degrees in Mechanical Engineer-ing from Worcester Polytechnic Institute(WPI), Worcester, Massachusetts, in1992, and 1999, respectively. He is cur-rently Assistant Professor of Mechani-cal Engineering at WPI working in thefield of NanoEngineering, Science, and

Technology (NEST). His professional interests include: combi-nation of modeling and simulation (CAD/CAE) with quantita-tive optical techniques, fiber optics, opto-electronic holography,nondestructive testing, image processing, numerical analysis,materials characterization, and optimization of mechanical andelectro-mechanical components.

Emily J. Pryputniewicz received herBachelor of Science and Master of Sci-ence Degrees in Mechanical Engineer-ing from Worcester Polytechnic Institute(WPI) in Worcester, Massachusetts, in1999, and 2000, respectively. While anundergraduate, she was named NASAScholar three times and conducted herundergraduate research on developmentof optical methodology for shape mea-surements at NASA-Langley facilities.

Her graduate research was conducted at the Sandia National Labo-




36

ratories, Albuquerque, New Mexico, where she concentrated onthe studies of MicroElectroMechanical Systems (MEMS). Cur-rently, she is continuing her work on MEMS as a Research Asso-ciate at the Institute for Defense Analysis, Alexandria, Virginia.

Gordon C. Brown holds a Bachelorof Science Degree in Applied Geophys-ics from Michigan Technological Uni-versity, and Master of Science and Ph.D.Degrees in Mechanical Engineeringfrom Worcester Polytechnic Institute(WPI), Worcester, Massachusetts, re-ceived in 1994 and 1999, respectively.From 1999 to 2000 he was AssistantProfessor at WPI and conducted researchin holographic interferometry tech-

niques for nondestructive testing ofMicroElectroMechanicalSystems (MEMS) at the Center for Ho-lographic Studies and Laser micro-mechaTronics (CHSLT).Currently, he is continuing to work on MEMS at the Corning-Lasertron in Bedford, Massachusetts.

Paul Galambos received his BSMEDegree from Bradley University in 1982,MSME Degree from The University ofTexas at Austin in 1987, and his Ph.D.Degree in Mechanical Engineering fromthe University of Washington in Seattle,Washington in 1998. In between peri-ods of education he has worked at Cat-erpillar Tractor Co. as a design engineer,Texas Instruments as an electronicspackaging engineer, and Lockheed/Mar-

tin as an aerospace engineer on the NASP program. He is cur-rently a staff member at Sandia National Laboratories in Albu-querque, New Mexico, working in the area of microsystems de-velopment, where he is conducting research in the area ofmicrofluidics with special emphasis on developing microfluidicsystems for biological applications.

ACES Characterization of Surface Micromachined Microfluidic Devices

Documents

Transcript of ACES Characterization of Surface Micromachined Microfluidic Devices