Rochester Institute of Technology

RIT Scholar Works Presentations and other scholarship

2007

Design and fabrication of microactuators and sensors for MEMS Ivan Puchades Robert Pearson Lynn Fuller Sara Gottermeier

Follow this and additional works at: http://scholarworks.rit.edu/other Recommended Citation Puchades, Ivan; Pearson, Robert; Fuller, Lynn; and Gottermeier, Sara, "Design and fabrication of microactuators and sensors for MEMS" (2007). Accessed from http://scholarworks.rit.edu/other/652

This Conference Proceeding is brought to you for free and open access by RIT Scholar Works. It has been accepted for inclusion in Presentations and other scholarship by an authorized administrator of RIT Scholar Works. For more information, please contact [email protected].

Copyright 2007 IEEE. Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must be obtained from the IEEE.

38

Design and Fabrication of Microactuators and Sensors for MEMS Ivan Puchades, 'Robert Pearson, 'Lynn F. Fuller, 'Sara Gottermeier and 2Sergey E. Lyshevski 'Department of Microelectronics Engineering

2Department of Electrical Engineering Rochester Institute of Technology, Rochester, NY 14623, USA E-mail: Sergey.Lyshevskigmail.rit.edu or seleeegrit.edu Web: www.rit.edu/-seleee Abstract - This paper reports the results for various microelectromechanical systems, devices and structures fabricated using bulk and surface micromachined processes. These microelectromechanical systems (MEMS) are designed and fabricated at the Semiconductor MicroFabrication Facility Laboratory at Rochester Institute of Technology. The microactuators and sensors are designed and fabricated for proof-of-concept lab-on-a-chip systems. The experimental results, which include testing, microof characterization evaluation and electromechanical actuators and sensors, are reported.

Keywords - ICs, MEMS, microactuator, sensor

I. INTRODUCTION

Recent advancements and developments in micro-electromechanical and microelectronics technologies have lead to the miniaturization and integration of MEMS devices in a great number of applications [1, 2]. In particular, lab-on-a-chip systems have greatly benefited from this technology evolution. Lab-on-a-chip solutions result in complex MEMS

that contain actuators, sensors, power sources, and ICs. The

integration of these many different components and devices result in significant integration challenges. A key challenge to the successful development of lab-on-a-chip MEMS is the coherent design and integration of all these components. The integrated design and fabrication of the actuators and sensor arrays is presented in this paper. The described design and fabrication taxonomy results in optimization, soundness and unified fabrication of close-loop controlled thermally actuated microfluidic pumps, anemometric gas flow sensors, Seebeck effect temperature sensors, interdigitated chemical sensors and light sensors. The proposed MEMS fabrication also allows for the integration of ICs.

DaGricaMion

CO F DIAGRAM II. FUNCTIONAL

OneIofIthe.ke

lab-ona One of the key challenges of lab-on-achallenges of the fabrication of chip system is the seamless integration of all of its components. These components are usually fabricated as stand-alone devices and combined into a lab-on-a-chip systm The degree of complexity of these systems is significant. For example, 4 different plastic/polymer based boards are compiled together to form a complete structure in [3]. Interconnection, compliance, alignment compatibilities

and other issues arise during this integration. The complexity, functionality, specified capabilities, integration, packaging and affordability requirements make the application of MEMS technology an ideal solution. The integrated fabrication of all the components of a labon-a-chip system not only simplifies processing and ensures affordability but also leads to superior performance and enhanced capabilities. Actuators, sensors, power sources and ICs can be integrated and fabricated on the same silicon substrate. A possible functional diagram for a lab-on-a-chip system is reported in Figure 1. Figures 1 and 2 depict the principle of operation of the proposed MEMS. A sample of liquid is pumped from a reservoir through a microfluidic channel to and adjacent thermal anemometer for closed-loop flow control before arriving to a mixing reservoir. The resulting product is pumped to final reservoirs where the pertinent analyses are performed. The output of these final sensors is easily extracted using the integrated ICs, such as an operational amplifier and a demultiplexer. Other MEMS can be designed applying the results reported.

III. ACTUATORS AND SENSORS

3.1. Thermally Actuated Microfluidic Pump A diaphragm type thermo-buckled aluminum/silicon multi-layer microactuator is integrated with a piezoresistive sensor. The actuator displaces the diaphragm of a microfluidic pump. The thermal expansion and different thermal expansion coefficients of the materials on the diaphragm may result in a force due to the thermal gradient, and the diaphragm is displaced. The equation describing the vertical displacement of a fixed two-layer circular plate, shown in Figure 3, due to a uniform temperature difference between the bottom and top surface is [4] (1) c, (ab - aa)(T - T )(tb + ta)(1 + Ve) rln a br

where cl is the coefficient which depends on the relative stiffness of layers; T is the temperature; To is the temperature at which the dahami lt i n i r h hra ofiin diaphragm is flat; a, and ab are the thermal coefficient expansions of the materials (for example, 22 ppm/C for a a almuan233p/°fosicn)tndtarthlye thickness; Ve is the effective Poisson's ratio of the composite membrane (Ve is ~0.3); r0 the radius of the heating element of the diaphragm; a is the membrane radius.

MEMSTECH'2007, May 23-26, 2007, Lviv-Polyana, UKRAINE

Authorized licensed use limited to: Rochester Institute of Technology. Downloaded on January 26, 2009 at 10:39 from IEEE Xplore. Restrictions apply.

39

OUTPUT

I FlOWu

Reservoir Pum

I ~~~~cmlp senso

_l

LL~LOCNIO

DEMUX 2

OSCILATO

Fig. 1. Lab-on-a-chip system includes micropumps, flow sensors, chemical sensors, temperature sensors and light intensity sensors integrated with ICs. Microfluidic channels, reservoirs and other structured can be fabricated using MEMS technology

monitored with the integrated sensor by utilizing the stress on the polsicilionpiezoresistors placed in a Wheatstone bridge at the edges of the diaphragm. The images are reported in Flowdirection section 5. The change in resistance of each of the polysilicon Thra resistors is related to the piezoresistive coefficient. The Chiem sensor Micropimi _ anemometer resistance variation is approximated as (3) AR=.cr Fig. 2. Cross-sectional view of the proposed microsystem using one where wT is the piezoresistive coefficient of the material; ar is the stress. substrate to fabricate all actuators and sensors Due to this Wheatstone bridge configuration, the thermal effect due to the heating of the polysilicon resistors on the A I¢a diaphragm is minimized. A microfluidic pump design is materiala ~~ _

tb ~¢~~~~~~~~~~~documented Fig.3. Circular multi-layer structure used to analyze the vertical deflection due to a temperature gradient. This equation does not apply to the case of the W microactuator under study since it does not take into account the expansion against the stationary silicon substrate._ However, it can be examined to determine that for maximum deflection the difference between the coefficients of thermal expansion of the two materials must be sufficiently large [8]. |iES The deflection of a square diaphragm is estimated as [9]

where y is the deflection at the center; p is the applied_ pressure; a is a length of the square diaphragm; v is the in-l plane Poisson ratio; E is the Young modulus (for silicon, E=1 .9x 01ol1 N/in2); h is the diaphragm thickness. Equaion 2)dscrbes he rlatinshp beweenpresure and vertical displacement of a square diaphragm, which can be used to determine the volumetric displacement of the micropump. The displacement and volumetric changes can be

in sectionl 5. The size of the diaphragm, the thickness of the diaphragm and the resistance of the heating coil can be optimized for a required volume displacement, pumping profile, etc. _

_

|

_ _l"

r

4M

_

Fig. 4. Top view of a thermally actuated micropump. The metal coi heats the surface of the diaphragm. The different thermal expansion results in avertical displacement. Polysilicon piezoresistors are deposited at the edge of the diaphragm to measure the displacement.

MEMSTECH'2007, May 23-26, 2007, Lviv-Polyana, UKRAINE

Authorized licensed use limited to: Rochester Institute of Technology. Downloaded on January 26, 2009 at 10:39 from IEEE Xplore. Restrictions apply.

40 3.2. ThermalActuators. Fundamentals Thermal diaphragm microactuators are utilized in inkjet printers. When the diaphragm is heated, thermally induced compressive stress causes the diaphragm to buckle rapidly and the diaphragm deflects toward the nozzle plate. The deflection increases the pressure in the ink chamber, and an ink droplet is ejected through the nozzle. The diaphragm is heated with a current of -0.5 A applying A10 V at 5 kHz [10]. The baseline equations for thermal actuators are examined. The generated heat per unit volume is -

Q(r>=pJ,

(4)

where p is the specific electric resistivity which is a function of temperature T, p(T)=po(I+b1T+b2I2+...); Po is the specific electric resistivity at zero temperature; bi are the temperature coefficients; J is the current density. The heart transfer rate q from surface area A can be estimated using the Nusselt number. The partial differential equations that describe the heat transfer are well known [11]. For homogeneous materials, the transient heat conduction equation (5) 8T,(t,r) (x2D77(t,r), reR, t>O

where a is the thermal expansion coefficient; AT is the temperature charge. The buckling occurs if S.S, due to the heating. For S=S, and S>SC, there are minima that correspond to the stable (steady-state) state of the stressed membrane. The force developed is given as F(r) =-VW,. (9) The membrane deflection is described by the partial differential equations, which are solved in the MATLAB environment. The experimental data that provides the

deflection of the membrane as measured using the sensors

reported in this paper is depicted in Figure 5. The results indicate the differential equations reported can be applied to describe the steady-state and dynamic behavior.

+ -

are solved using the boundary conditions applying high.........

performance software such as MATLABor ANSYS [11]. Here,

--------------Fig 5 Membrane displacement dynamics for a step pressure change ez2 er2 ex2 3.3 Anemometric Flow Sensor (6) Anemometric flow sensors are MEMS thermal sensors where a1t iS the thermal diffusivity of the solid. For the thermal for measuring liquid flow rates [5]. Their basic principle of multi-layer actuators, the application of the partial differential operation is based on the convective heat transfer as described equations is complicated due to various layers, complex by V21L geometry, nonuniformity, parameter variations (which are ' (10) Rthermal, Rthermal = 1 T-T + constant for bulk media), etc. Therefore, we perform focused R C.A studies applying sound assumptions which do not change overall quantitative and qualitative results. Those analyses are where T is the final temperature, T, the initial temperature, V centered on the basic-applied-experimental-and-technology is the voltage applied, R is the resistance of the polysilicon resistor and Rthe~a1 is the thermal resistance to ambient. To co-design with a focus not only on the fundamental research but also on fabrication as well as proof-of-concept test-bed calculate the thermal resistance to ambient, we use the testing, characterization and evaluation. The qualitative following parameters: C is the thermal conductivity of the material, L the length to ambient, and A the cross-sectional equations are derived to describer the diaphragm deflection force and temperature. The applied voltage varies to deflect area of the path to ambient. The polysilicon resistor is built directly on the bottom of the membrane. The above listed quantities and variables can in the close roximit to the fluid flow. The channel Or 2a23the . be measured or observed. For a axa membrane with a y thickness h, assuming the uniform compression t, the total polyslicon reslistance changes near-liearly with temperature strai erg is within the operating envelope. The sensor iscalibrated to 2tiobtain the sensor's temperature coefficient of resistance. 5 5qu l m Eh eo Using a constant current power supply, the output voltage 1w - S W -cYo l s p n 2 Sw) across the res stor is measured for temp erature calibration. ) 2y sohn) 21s 'stu () where Yo is the deflection at the center;Su is the critical stress Additional calibration is needed to obtain the heater resistance eisthe Young's N/im2foreSlicon)-visthe changesi asnd a function of fluid flow. To avoid heat loss Poisson's ratio (c0.3 for Silicon); cl and c2 are the constants through the substrate and improve sensitivity, the sensor is built on a thin silicon membrane. cts3 and c2 c8. The compressive stress is 3.4. Seebek Effect Temperature Sensor (1 + v)raE When two dissimilar materials form a junction a voltage (8) St lr 2\ AT, polyimight be generated if the junction is at a different temperature thickness h assuming thetotal v theuniformcompression than the other end of the conductors or cold junction. This is in the Cartesian coordinates

2T (t,r)-277(t,x,y,z)

82T(t,x,y,z) + ey2

7,(t,x,y,z)

'8?

1-iv2

modulusi(.9x0ld

MEMSTECH'2007, May 23-26, 2007, Lviv-Polyana, UKRAINE

Authorized licensed use limited to: Rochester Institute of Technology. Downloaded on January 26, 2009 at 10:39 from IEEE Xplore. Restrictions apply.

41

the principle used in thermocouples. The expected voltage change is related to the Seebeck coefficients (a, and a2) of the materials and can be calculated using the following

equation [6] A V=a (T

T-Tt) + a

(l

The Seebeck coefficient for a n-type Polysilicon is -100 gV/0K, while for aluminum, it is 4.2 gV/0K. By designing the cold junction to reside over the silicon bulk area outside of the thin diaphragm, an accurate reading of the temperature of the

sample can be obtained.

IV. BULK MICROMACHINED PROCESS

The fabrication process starts with a bare double-sidepolished n-type silicon wafer. A silicon oxide is grown and used as a masking layer for the P+ spin-on-dopant process. This silicon oxide is patterned and etched to create a low resistivity boron-doped diffused silicon layer for the p-n diode/light intensity sensor. After this, a pad silicon oxide is thermally grown and silicon nitride is deposited using a lowpressure chemical vapor deposition (LPCVD) process. The silicon wafers caused by the nitride. The silicon nitride layer is used as a protection layer against the KOH etch. Before continuing with the KOH etching, the wafer is processed through the backside photolithography step. The patterned silicon nitride and oxide are then removed from the backside of the wafers by dry SF6 and HF etch respectively, Backside etch in KOH is a critical step in the fabrication of thin flexible silicon diaphragm particularly in applications that require large diaphragm deflections. In order to achieve a specified thickness, the etch rate needs to be calculated. This

is particularly important when process conditions such as

......

3

i

_ 11111111111111 1111111111111111. lilll_ 1111111 Fig. 7.SEM cross-section of the 351tm Si diaphragm top of a via-patterned insulating layer of io,oooA of LTO allowing for a more compact and complex design. Interdigitated chemical sensors are built during these final steps ensuring their exposure and reaction to chemical compounds to be applied to the topmost layer of the system.

V. RESULTS -THERMAL ACTUATORS AND

KOH bath temperature and original silicon wafer impurity SENSORS Several versions of the square thermally actuated concentrations are not tightly controlled. Diaphragm thicknesses in the range from 20 to 40 pm have been achieved microfluidic actuator pump have been fabricated and tested. with this process. The optimal thickness of the diaphragm The diaphragm dimensions are varied from 2x2mm to 3x3Tmh the diaphragm thickness from 60 to 20 pm and the needs to be found to achieve controlled pumping (pulse, continuous and hybrid) and to meet robustness, uniformity, Al coil metal resistance from 2 to 300 ohm. The vertical r A 35 pm thermal insulation and flexibility requirements. displacement can be in-situ measured with the polysilicon thickness ensures wafer damage tolerance during the piezoresistors Wheatstone bridge fabricated on the diaphragm remaining fabrication processes as well as stiffness, thermal itself Figures 8.a and 8.b show two different designs, which insulation and robustness during pumping operation. After the have been used to characterize the behavior of these thermal silicon diaphragm is created, both the protective silicon nitride actuators for micropump applications. The Al coil resistance and the remaining pad silicon oxide are removed. Figure 6 is 2.2 and 250 ohm respectively. Wheaston brde the piezoresistors voltage thcnesoees ensue wafer dhamagoe duigtepezoresistors 7 reports Figures 9.a and 9.b report ariaeohedaprg shows an image of the backside of the toleac wafer. Figure across-section ofthe silicon diaphragm. output as a function of the current the coils. The heating Polysilicon is then deposited via LPCVD obeentop oan power of the device in Figure 8.a is 352 mW, which corresponds to an increase temperature of11rCo while insulating oxide telayer. n for single-wire .b inThepoertthe pizoemsistors vlage Fighrtmaures show an mageof baksieThermocouples, ofthe afer reot bpidge putlze, Figure 8.b is Pinhr.6 W and a temperature increase is 137 °C. anemomedster aond the Wchieatste pizontroles Figref fabrication a the much larger the second design results inwith remaining procesesasellasstTherfore, draw A low temperature layer iS then mae deflecto andpmine n-type doped polystlicon. . The At teprtresand deposilted and contact openngs to poly and silicon are etched oac higher tempeature solution. After the contacts are etched, a first displaceet vo cn be austhe toi mee dif n ou an tHFin of aluminum is applications. layer of deposited and then metal 1sOOOA patterned to make the electrical connections. A second depOsiA aluminum metal layer is deposited and patterned on

MEMSTECH'2007, May 23-26, 2007, Lviv-Polyana, UKRAINE

Authorized licensed use limited to: Rochester Institute of Technology. Downloaded on January 26, 2009 at 10:39 from IEEE Xplore. Restrictions apply.

42

Amplitude vs Frequency

E 1

(b)

(a)

Fig. 8. Low resistance and high resistance thermal actuators with integrated piezoresistors sensors as a Wheatstone bridge configuration. Low resistance design

1.200

1.000

lo111.000 4

_ 0.800

0.400

E0.600 0

200

400

600

0.400

0

Current (mAmps)

200

400

1000

in thermally actuated micropump.

0 \800

|E 0.600

100

Fig. 11. Frequency response of a high resistance 3x3 mm diaphragm

High resistance design

1.200

10

Frequency (Hz)

600

An image of the single-wire anemometer sensor is shown in Figure 12. This device has a built-in heater allowing the device characterization capabilities. The results of the singlewire anemometer flow sensor are presented in Figure 13. The purpose of the flow sensor is to monitor the volume displacement of the microfluidic pump. Together with the piezoresistors output of the pump, the system can self-tested and reconfigured the pump's volumetric displacement in order to obtain the needed flow rate for the application.

Current (mAmps)

(a)

(b) bridge voltage output in low and high resistance microactuators. The high resistance actuator results in much larger vertical displacement. Fig. 9. Wheatsone

The operation frequency testing is performed. For the high resistance device, presented in Figure 8.b, the results are reported below. A square pulse of 30V (lOOmA) with a 50 00 duty cycle is applied to the Al coils. Figure 10 shows the results if 30 V, 200 Hz pulses are applied to the heating coil,

and a 25 ohm resistor is connected in series to obtain the Fig. 12. Single wire polysicilion anemometer with an integrated current. We confirm the thermal effct anddissipationasthe o clbain current flows through the coil. Correspondingly, we confirmdfue hae the thermal actuation mechanism of the microactuator. Figure I1I shows that the frequency response of the piezoresistive Flow sensor calibration 350output remains constant up to 100OHz. .'H1

-

......

-

l :F___'i |___

_

l-l

a)

=330

|

I rl-

310

I

Z Co

-

290 270

250

0

10

20 30 Heater Voltage

40

Fig. 13. Anemometer resistance versus heater voltage. | Fig. 10. Voltage drop in the coil: 30 V voltage pulses are applied at 200 Hz.

~~The thermocouple sensor is shown in Figure 14. Integrated heaters ensure testing and calibration. The voltage output response of the thermopile is obtained in terms of

MEMSTECH'2007, May 23-26, 2007, Lviv-Polyana, UKRAINE

Authorized licensed use limited to: Rochester Institute of Technology. Downloaded on January 26, 2009 at 10:39 from IEEE Xplore. Restrictions apply.

43 power through the heater as reported in Figure 15. Although the voltage change response is small, it is linear, which allows the use of an operational amplifier to accurately measure small temperature changes.

Fig. 16. ITD chemical sensor coated with a chemical compound

Fig. 14.Polysilicon and aluminum thermocouple array to monitor temperature. This sensor has diffused resistor for temperature

calibration

Measured

0.02

0.015 >

s

0.01

Fig.| 17. ITD chemical sensor response to exposure. When the sensor

|

is coated with a specific chemical compound, the resistance is low. As it exposed to a certain ambient, the resistance increases. diode

-

| 0.005

Increased ligt

0

0

0.1

0.2 0.3 Power (W)

0.4

Fig. 15. Delta voltage output of the thermocouple array in response to increased power dissipated through the calibration heater

Chemical sensors can be designed as interdigitated metal lines coated with a material that changes the resistance when exposed to certain chemicals. An image of the fabricated interdigitated chemical sensor is shown in Figure 16. The sensor is covered with a chemical compound that results in a low resistance being measured between the two metal terminals. Figure 17 shows that when the sensor is exposed to a certain ambient, the resistance increases. As it is removed, resistance decreases to an initial value. Different chemical compounds, which are reactive to different agents, can be used in this design as reported in [7]. The depleted space charge region of a p-n junction collects the electron-hole pairs generated by the incoming radiation. The generated current is proportional to the light intensity and can be measured. The I-V characteristics of the light intensity diode are shown in Figure 18. An increase in the reverse bias current is seen as the light intensity is increased.

intensity