7/29/2019 the myth of fight

1/4

SU-8 BASED PIEZORESISTI VE M ECHANI CAL SENSORJ . Thaysen, A.D. Yalqinkaya2,R.K. Vestergaard2,S.J ensen2,M.W. Mortensen2,P.Vettiger3 and

A. Menon2CANTION A/S and2MikroelektronikCentret, Technical Universityof Denmark, bldg. 345east, DK -2800K gs. Lyngby, Denmark, 31BM Research Laboratory,CH-8803Riischlikon, Switzerland.ABSTRACT

We present the first SU-8 based piezoresistivemechanical sensor. Conventionally, silicon has been usedas a piezoresistive material due to its high gauge factor andthereby high sensitivity to strain changes in a sensor. Byusing the fact that SU-8 is much softer than silicon andthat a gold resistor is easily incorporated in SU-8, we haveproven that a SU-8 based cantilever sensor is almost assensitive to stress changes as the silicon piezoresistivecantilever. We demonstrate the chip fabrication, andcharacterization with respect to sensitivity, noise anddevice failure.INTRODUCTION

The use of the SU-8 polymer within theMEMSfield has been exponentially growing during the lastcouple of years. The fact that SU-8 is very chemicallyresistant makes it possible for the use as a componentmaterial. Due to its ability of defining layers withthicknesses between1pm and 1 mm with high aspect ratio(>20) [1], SU-8 has been a popular and cheap alternativeto silicon for the fabricationof passive components. Suchcomponents include microchannels, micromolds forelectroplating or masters for hot embossing. Passive SU-8based atomic force microscopy ( AFM) cantilevers havealso been demonstrated [2].

Since silicon exhibits very good mechanicalbehaviors and also a very high piezoresistive coefficient,SU-8hasso far never been considered as an alternative asa sensor material with integrated readout. In this paper wepresent the first SU-8 based sensor with piezoresistivereadout which is a cheap and easy fabrication alternativecompared to silicon sensors. A cantilever withpiezoresistive readout is presentedasan example of suchasensor. Cantilevers with piezoresistive readout arenormally used as stress sensor, e.g. AFM probes andsurface stress sensors for detection of bio-molecularinteractions.THEORY

In order to illustrate the sensitivity of the SU-basedpiezoresistive cantilever, it is compared to the sensitivityof a piezoresistive silicon cantilever. In this example thesurfaces stress sensitivity s compared for the two differentsensors.

When molecules bind to the surface of acantilever, the surface stresso, changes due to molecularinteractions. This stress change can then be picked-up inthe integrated piezoresistor.A simple expression for thesensitivity can be obtained by assuming that the cantileveronly consists of one material and an infinitely thin resistoris placed on top of the cantilever. The relative change inresistance can be writtenas:

whereK isthe gauge factor,E is Youngs modulus andhis the thickness of the cantilever.In our sensor, a thin gold film is used as thepiezoresistor. Gold has a low gauge factor (&,=2)compared to silicon(Ksi=140) and is therefore consideredinferior to silicon asapiezoresistive sensor material. Fromequation (1) it is seen that theWE actually determines thestress sensitivity of the cantilever for the same thickness.SinceSU-8 has a Youngs modulus of 5 GPa and siliconhas a Youngs modulus of 180GPa, the ratios becomes(K/E)~=0.8GPa- and ( I U E )S~- ~~~~=O. ~Pa-, which isonly a factor of 2 in sensitivity in favor of silicon. Thesensitivity of theSU-8 based piezoresistive cantilever canin principle be enhanced by integrating a piezoresistormaterial with even higher gauge factor. It is for examplepossibleto integratea sputtered silicon piezoresistor withagauge factor of about 20. But in order to use Youngssmodulus for SU-8 in theWE relation, the stiffnessof thepiezoresistor should be neglectable compared to the SU-8cantilever. This can only be obtained by reducing thethickness of the poly silicon resistor which increases thenoise significantly and thereby reducing the signal to noiseratio.

Contact Au wire Au resistorTopSU-8 layer

/ su-asdostrate

Figure1: Schematicof thechipdesign.

Jacob Thaysen, CANTIONNS,DTU, CAT, bldg. 347, DK-2800 Lyngby, email:jt@cantion.com, tlf. +45 4525 6409

0-7803-7185-2/02/$10.0002002IEEE 320

mailto:jt@cantion.commailto:jt@cantion.com

7/29/2019 the myth of fight

2/4

DESIGN

Si wafer

The design of the cantilever chip is shownin figure1.Ascan be seen from the figure that the chip consists of twocantilevers with integrated resistors and two resistors onthe substrate forming a whole Wheatstone bridge. Theadvantage of this design is that one of the cantilevers canbe used as a measurement cantilever, while the other isused as common-mode rejection filter. The parameters forthe fabricated cantilevers are shown in table 1.

(5OAdOOAROOA)

Table1: Design parameter:Parameter I Value I UnitCantilever length 200 pmCantilever width 100 pmCantilever Thickness 7.3 pmSpring constant 7 N/mResonant freauencv 49 kHz

Optical pictures of the fabricated chip can be seenin figure 2. In figure 2a, thetwo cantilevers can be seen.Figure 2b shows a close-up of one of the cantilevers. Themeander-like resistor structure is clearly seen in thepicture.

fast etching sacrificial layer, developed by G. Genolet [3].A thin layer of SU-8 is spun on the wafer and patterned asupper cantilever layer (figure 3b).A 1 pm thick gold layeris deposited on top of the wafer. The conductor mask istransferred to the wafer by standardphotoresistlphotolithograph? and the conductors areetched (figure 3c). A 400A thick gold layer is depositedand the resistors are defined by the same procedure asabove (figure 3d). The metal is completely encapsulated inSU-8by depositing and patterning ofa5.8 pm thick SU-8layer for the lower SU-8 cantilever layer (figure 3e). A350 pm thick SU-8 layeris spun on the wafer an patternedas the chip substrate (figure 30. The chp is finallyreleased by etching of the sacrificial layer (figure 3g).

LAu resistorsSU-8 bottom layer(5 eum)- U-8 suhstrate layer(35OumFigure2: Optical images of the chip. Image a) showsthe two cantilevers inserted in the on-chip Wheatstonebridge, while image b) shows a close-up of thecantilever with integrated meander-typeresistor.

Figure 3: Process sequence for the fabrication of theSU-8cantilever.

FABRICATIONThe SU-8 based cantilever with integratedpiezoresistive readout is fabricated on a silicon substrate,which isonly used in order to be able to handle the chipsduring the processing. First, a Cr/Au/Cr layer is depositedon the silicon wafer see figure 3a. This is used as a very

CHARACTERIZATIONSEM micrograph of the finished chip is seen infigure4. Figure 4a shows the cantilevers seen from below.The cantilevers are seen to be completely straight. Figure4b shows the cantilevers seen from above. It is notpossible to see the resistors or the conductors on thismicrograph.

321

7/29/2019 the myth of fight

3/4

Figure4: SEM micrographs of the cantilever chip, a)seen from below and b) seen from the top.

cantilever0.3.10-6eflection sensitivityInmP

The deflection sensitivity of the piezoresistiveSU-8 cantilevers has been measured by observing therelative change in resistance as a function of the cantileverdeflection. The result is shown in figure5. It is seen that astraight line can be obtained from the measurement. Thisstates that the deformationspurely elastic.(optimized)4.8.10'6

Deflection(pi)Figure5: Deflection sensitivity measurement graph

4inimum detectabledeflection

The deflection sensitivity can be found from theslope of the straight line to %lz=O.3 ppdnm, whichyields a gauge factor of K=4. The reason that the gaugefactor is higher than expected is probably due to differentbehavior of the very thin gold resistor thana bulk goldresistor.

0.4

~----tlohnsonnoisem - - 4 ovp-94.w

3.104urface stresssensitivity [N/m]-'

-.. , -- I

1.10.~

1 10 100Figure 6: The graph shows the electrical noise as afunctionof frequency for different supply voltage.

Frequency Hz)

The minimum detectable deflection or minimumdetectable surface stress is defined by the noise in thesystem. Since the vibrational noise is considerably lowerthan the electrical noise sources in this resistor setup, onlythe Johnson noise and the 11f noise is considered here. Infigure6, the noise is measuredas a function of frequencyfor different input voltages. It is seen from the graph thatthe llf noise is very low withaknee frequency of about10Hz for a Wheatstone bridge supply voltage of 4.5V. It isseen from the graph the measured Johnson noise floor isaboutafactor of 2 higher than expected.Table 2: Performanceof theSU-8based piezoresistivecantilever compared to a piezoresistive siliconcantilever.r ~ Parameter 1 SU-8 I Sicantilever I

1 1.104 I 2.10-5 1inimum detectablesurface stressTNhlFrom the above measurements it is possible to listthe performance of the SU-8 based piezoresistivecantilever. The performance is listed in table 2 withrespect to deflection sensitivity, minimum detectable

deflection, surface stress sensitivity and minimumdetectable surface stress. Furthermore, the performance iscompared to an optimized silicon piezoresistive cantilever.It is seen from the table that the minimum detectabledeflection is 10times better for the silicon cantilever, butonly 5 times better for the minimum detectable surfacestress, nevertheless the SU-8 based piezoresistivecantilever can be used as a surface stress bio-chemicalsensor, since the change in surface stress due to molecular

322

7/29/2019 the myth of fight

4/4

interactions on a cantilever surface is normally in the orderof - 1 N/m [4,5]. Reducing the thickness of thecantilever can increase the surface stress performance evenfurther. As seen from equation 1, the sensitivity isinversely proportional with the thickness. With the giventechnology it is possible to decrease the cantileverthickness a factor of 2 and thereby decrease the minimumdetectable surface stress with a factor of 2.Electron migration of the gold resistor has turnedout to limit the supply voltage. As seen from equation 2the mean time to failure(MTF) due to electron migrationdecreases as the squareof the current density(i).

where Q is the diffusion activation energy and T is thetemperature. Since the gold resistor has a very small cross-sectional area this effect quickly becomes dominant factorfor the working of the device.1.80, I

t 1

1.75i2598 1.70v)

1.650 40 80 120 160lime (sec.)Figure7: The graph shows the resistor value becomesunstable for a supply voltage more than 1 V over the

resistor.

Figure 8: In both images the lower cantilever isexposed to a voltage drop. In a) the cantilever isexposed to 1V and no plastic deformation is obtainedand b) the cantilever is exposed to 2 V at which it isdeformed.Figure 7 shows two plot of the change inresistance value as a function of time for different voltagedrops over a gold resistor. The resistor value becomesunstable for supply voltages more than1V. As seen in thefigure, the resistor value for a voltage drop of 2 V isunstable due to the deformation of the cantilever. After thedeformation is completed the resistor value stabilizes. Forsupply voltages higher than 3V the resistor burns within a

few seconds. The reason for the instability of theresistance is not only due to electron migration, but alsodue to a plastic deformation of the cantilever. Figure8shows the plastic deformation of the cantilever fordifferent supply voltages. It can be seen that nodeformation is obtained for a supply voltage of 1V whilethe cantilever is severely bended due to a plasticdeformation for a supply voltage of 2 V. The reason forthe plastic deformationisprobably due a significant self-heating of the resistor. Since the SU-8 polymer is a verypoor heat conductor, the resistor self-heating has probablyheated the SU-8 non-uniformly through the cantilever,which has lead to the bending of the cantilever. From thisexperiment it can be concluded that the maximum voltagedrop over the cantilever resistor should be 1 V,corresponding to a Wheatstone bridge supply voltage of 2V.

CONCLUSIONSWe have demonstrated the design, fabrication and

performance of the first SU-8 based piezoresistivecantilever. The fabrication of the sensor is very cheap andeasy compared to conventional silicon fabricationtechnology, and the use of SU-8 opens up for a whole newdesign space. The theory described why the surface stresssensitive cantilever potentially is as sensitive as a siliconcantilever, and the characterization of the fabricatedcantilever supports this theory. The measured gauge factorwas a factor of 2 higher than expected. For the currentdesign, the minimum detectable surface stress is still afactor of 5higher than for the optimized silicon cantilever,but is still sensitive enough for the use in many bio-chemical sensor applications.ACKNOWLEDGEMENT

We would like to thank Ole Hansen and AnjaBoisen from Mikroelektronik Centret for fruitfuldiscussions and many insightful inputs.REFERENCES

[l]H. Lorenz et al.. SU-8: a low-cost negative resist forMEMS. J . Micromech. Microeng., 7: pp.121-124,1997.G. Genolet et al., Soft, entirely photoplastic probesfor scanning force microscopy, Review of ScientijicInstruments, Vol. 70 (5),pp. 2398-401,1999.G. Genolet, New photoplastic fabrication techniquesand devices based on high aspect ratio photoresist,Ph.d. thesis fr om Swiss Federal Institute of TechnologyLausanne, 2001.[4]G. Wu et al., Origin of nanomechanical cantilevermotion generated from biomolecular interactions,Proceedings o the National Academy of Sciences ofthe USA,pp.1560-1564,2001.[5] J . Thaysen et al., A microcantilever-based detectionprinciple for pTAS, Proceedings of pTAS 2001,Monterey, USA, pp. 447-449,2001.

323