Sensors and Actuators A 160 (2010) 3541

Contents lists available at ScienceDirect

Sensors and Actuators A: Physical

journa l homepage: www.e lsev ier .co

A hollo se

P.K. KinnGE Sensing & In

a r t i c l

Article history:Received 15 DReceived in reAccepted 12 MAvailable onlin

Keywords:Pressure sensoElectrochemicHollow structuBossDiaphragmSilicon

or proical efocusof thprocsuitabnsore preearity

1. Introdu

Piezoreswheatstone bridge of piezoresistors fabricated on a square silicondiaphragm. The sensitivity of these sensors will be inversely pro-portional to the square of the diaphragm thickness, and directlyproportional to diaphragm area [1]. To fabricate higher sensitiv-ity devices the designer is forced to increase die size or reducediaphragmtowards rederally to redsensitivity.ically fabricsilicon, andcareful timimonitor thevariation frwill ultimacally diaphrmanufacturare a numbmatically stthickness huse of silico[4], or an el

CorresponLeicestershire,

E-mail add

al ets thag th

turing of the diaphragms to improve other device performancecharacteristics such as output linearity.

As high sensor linearity is often a very attractive sensor char-acteristic, at silicon diaphragms are generally modied with theaddition of a lump or boss structure to stiffen the centre of the

0924-4247/$ doi:10.1016/j.thickness. Cost constraints often lead the designerucing die size. Therefore the approach taken is gen-uce diaphragm thickness in order to meet the requiredSilicon diaphragms for MEMS pressure sensors are typ-ated by wet anisotropic etching of a cavity into bulkto achieve a required diaphragm thickness requires

ng of this etch. Even if great care is taken to continuallyetch in order to achieve the desired thickness, process

om wafer to wafer, and from die to die within a wafertely dictate the minimum practicable thickness. Typi-agms thinner than 40m prove challenging for manying environments [2]. To overcome this problem thereer of etch-stop methods that may be employed to auto-op or limit the anisotropic etching once the requiredas been reached. Examples of such techniques are, then on insulator wafers [3], the boron etch-stop processectrochemical etch-stop [5,2]. In this work the electro-

ding author at: GE Sensing, Silicon Engineering, Fir Tree Lane, Groby,LE60FH, UK. Tel.: +44 1162317507.ress: Peter.kinnell@ge.com (P.K. Kinnell).

diaphragm. Fig. 1 shows a micrograph of a typical pressure sen-sor diaphragm with a central stiffening boss, the cross-sectionmarked on the micrograph can be seen as a schematic in Fig. 3b.These features improve linearity by limiting strain stiffening of thediaphragm, which is a signicant cause of output non-linearity [6].The addition of such stiffening bosses is often done at the expenseof other device parameters such as die size. Fabricating the boss onthe diaphragmgenerally increases die size, and the additionalmassthat is suspended will tend to cause increased acceleration sensi-tivity. This acceleration sensitivity becomes especially critical ashigher sensitivity pressure sensors are fabricated. This is becausethe diaphragm that is supporting the boss becomes increasinglyexible, thus any inertial loads imposed on the boss will result ina greater deection of the diaphragm and thus be seen as a moresignicant proportion of the sensor output.

To address these issues this work presents a novel fabricationroute that allows for a wide variety of hollow structures to be fab-ricated on the surface of a thin silicon diaphragm. These structuresmaybeused, in effect, as hollowbosses that have considerable stiff-ness relative to the diaphragm, yet due to their hollow constructionadd very little mass to the diaphragm (see Fig. 3c). The diaphragmand hollow bosses are produced using an electrochemical etch-

see front matter 2010 Elsevier B.V. All rights reserved.sna.2010.03.024w stiffening structure for low-pressure

ell , J. King, M. Lester, R. Craddockspection Technologies, Fir Tree Lane, Groby, Leicestershire, UK

e i n f o

ecember 2009vised form 18 February 2010arch 2010e 18 March 2010

ral etchre

a b s t r a c t

This paper presents a novel process fcon diaphragms using an electrochemmethod are presented together withture. These demonstrate the integrityMEMS sensor applications. Using thisdemonstrated, and used to verify thenovel process allows for increased sesensor design and characterization arlinearity

36 P.K. Kinnell et al. / Sensors and Actuators A 160 (2010) 3541

Fig. 1. Microgcentre, the dia

stop technithin diaphrthese featuin terms ofinertial senresult frombosses. Forconstraintsically requithen etchedshaped bosEtching bosuse bossesadverse affleads to anby increasinlimited by tof wafer thi

In the fooped to fabthat in prinof generic hA specic ea piezoresisfabricationpressure seare presentsor and valmechanical

Fig. 2. Cross-smost basic forof diaphragmlump.

ross-slanes

hol

rderKOHrvieain pitabld boimpate (sviousubsop pelecfromingsurftilisesilicve aypeut cacessatcheself-sustaining under the cell potential, the integrity of theraph of an etched pressure sensor diaphragmwith a solid lump at thephragm is approximately 1.5mm square.

que and therefore this allows for controlled etching ofagms that are

P.K. Kinnell et al. / Sensors and Actuators A 160 (2010) 3541 37

Fig. 4. Process ow to create hollow boss s

Fig. 5

In this wduring a wehollow bosplete removlayer with t

Fig. 6. Chart shical etch-stop

lowboss str80m deepvery thin-wof the proclar interestand the wacates the pothat the wathese areas

ssesr messesnd ssyslow boto poolow bo(FIB) athe FIB. Diagram of electrochemical etch-stop equipment set up.

ay n-type structures were preserved to full integrityt anisotropic etch, evidenced by the fabrication of thes structure in Fig. 7. This micrograph shows that com-al of the p-type siliconwas obtained, leaving the n-typehe buried hollow boss structures bonded to it. The hol-

owing relationshipbetweencurrent andvoltage for theelectrochem-process.

milling suchdiaphragmis shown inor other strhole that wto measuremately 2.4slightly thictop of the brate of thepassivationthin-walledwork the neusing anisotive structuFor examplused to credepending

3. Test sen

The hollbossesmaytures may bwas designdescribed athis type ogiven by thelinearity dusensor wasrange of 20of approximpressure seis a challenbenet fromcontrol diapmise sensotructure.

ucture shown in Fig. 7, is approximately 400msquare,with a wall thickness of 3m. These relatively largealled structures were seen as a good test of the abilityess to produce well-formed hollow bosses. Of particu-was the area of intersection between the diaphragm

lls of the hollow boss as shown in Fig. 7a, an arrow indi-int of interface between the boss and diaphragmnotingll thickness at this interface is approximately 3m. Ifare not fully bonded together then the use of the hol-as structural elements in sensorsmay be prohibited duechanical stability. The structural integrity of the hol-were analysed using a combination of focus ion beamcanning electron microscopy (SEM) techniques. Usingtem a section of the hollow boss was removed by ionthat the interfacebetween thewalls of theboss and the

could be observed. Amagnied image of this cut sectionFig. 7b, fromwhich it can be seen that there are no voidsuctural faults that may lead to poor performance. Theas cut in the at section at the top of the boss was usedthe top layer thickness. This was found to be approxi-m indicating the sidewalls at the base of the boss areker than the at top of the boss. It is expected that theoss will be slightly thinner that the base due to the etchsilicon under etch-stop conditions, which is set by theoxide etch rate. This structure demonstrates that largestructures may be created with this process. In thisgatives used to create the hollow bosses were createdtropic etching, hence the pyramidal shape. These nega-res can also be created using other etching techniques.e, isotropic etching or deep reactive ion etches may beate bosses with either rounded or vertical sidewalls,on the required application.

sor design

ow structures detailed above demonstrate that hollowbemanufactured, however, to determine if these struc-e used as part of MEMS sensors a low-pressure sensored and fabricated using hollow bosses and the processbove. A low-pressure sensor was specically chosen as

f device benets from the controlled thin diaphragmshollow boss process, as well as optimal sensitivity ande to the addition of hollow bosses. The low-pressuredesigned to have a target full-scale pressure in the30mbar. This is based on achieving an electrical outputately 20mV/V output at full-scale pressure. A low-

nsor was chosen to demonstrate this process, as thisging area of design space for pressure sensors. It would

the combination of an electrochemical etch-stop tohragm thickness with the hollow boss process to opti-

r performance.

38 P.K. Kinnell et al. / Sensors and Actuators A 160 (2010) 3541

Fig. 7. (a) Focused ion beam section cut from a hollow boss and (b) close-up of cut section that is indicated by the arrow shown in (a).

The pressure sensorwas designedwith a diaphragmof 1550msquare and 5m thick, onto which three hollow boss structures80m high were xed. The wall thickness of the hollow bosseswas designed to be approximately 10m; the overall geome-try of the dthe most bdiaphragmthree bossesor performeffective bebosses butthat whentrated in thillustrate thfrom pressuated usingthe bossesbosses andoping maxiFig. 9 indicated betweFig. 10, in tated betwethis type ofform stressthe positionThe resistoindicated opositions ofcorrespondshown.

Fig. 8. G

tress plot fromAnsys 10.0 nite element software after a pressure is appliedwer side of the diaphragm, the exaggerated deection plot illustrates howes remain rigid and stress is concentrated in the areas of diaphragmbetweenes.

well as ease of positioning of the piezoresistors, the threealso concentrate stress such that for a given diaphragmincreased sensor output is achieved with a reduced non-

ty. The exact size of the three bosseswas arrived at followingiaphragm can be seen in Fig. 8. As mentioned above,asic pressure sensor would comprise of a simple at(see the cross-section shown in Fig. 3a). In this works have been added to the diaphragm to improve sen-ance. Adding the three bosses to the diaphragm iscause it stiffens the diaphragm in the regions of thekeeps the diaphragm exible elsewhere. This meanspressure is applied to the diaphragm stress is concen-e areas that are kept exible between the bosses. Tois Fig. 9 shows an exaggerated deection plot resultingre applied to the lower side of the diaphragm (cre-

Ansys FEA software). From the plot it can be seen thatremain rigid with the areas of diaphragm between thethe frame taking up all the bending and therefore devel-mum tensile and compressive stresses. The arrows inate the areas of maximum and minimum stress situ-en the lumps. A close-up of this area is also shown inhis plot the areas of uniform stress that have been cre-en the bosses can be seen. This is a key advantage ofdiaphragm over a conventional at diaphragm. Uni-

es have thus been created between the bosses aidinging of piezoresistors to sense pressure-induced stress.rs would therefore be positioned in these regions asn the diagram shown in Fig. 11. In this diagram thethe resistors are shown relative to the bosses, and theing position of each resistor in the sensing bridge is also

Fig. 9. Sto the lothebossthe boss

Asbossesarea anlinearieometry of low-pressure test sensor with three hollow bosses.

Fig. 10. Closebetween the hthe hollow bo-up stress plot of the areas of maximum andminimum stress situatedollow mesas. The view is from below the diaphragm looking up withsses on the far side.

P.K. Kinnell et al. / Sensors and Actuators A 160 (2010) 3541 39

Fig. 11. Schem ith thealong with a d rs in

an iterativeanalysis mefor two comdiaphragm,The solid bsponds to toverall size1550m sqnoted thatthe centralnot been adtall, which abecome towdone usingof the diaphear static anpressure loboth sides otional pressto simulatecase of thewas modelinside the bcess used tthat introdand the beity is approwith similafor accelerais the leastthree timesthe solid boa three bosthe hollow

Table 1Comparison of

Device

Hollow bossSolid boss 38No boss at

is b.

ract

fullyfor

essuvityusinof sely, spattacto thm gThe mce odicaollowerstab

fromatic top view of the test sensor showing the three hollow bosses on a diaphragmwiagram of the wheatstone bridge used and the corresponding position of the resisto

optimisation process carried out using nite elementthods. As a basic comparison the estimated outputparative devices was considered, these were a at

andadiaphragmwithasolidbossof380m,seeFig. 12.oss was constrained to be 380m high, which corre-he thickness of a typically available silicon wafer. Theand thickness of the diaphragms was kept the same atuare and 5m thick. For the solid boss die it must beonly one boss is used that has the same base length asboss on the hollow boss device. The side bosses haveded because these bosses are constrained to be 380ms a result of the anisotropic etching processmeans theyide to t them on the diaphragm. The comparisonwasAnsys FEA software to run a non-linear static analysisragms under the inuence of a pressure load, and a lin-alysis under the inuence of an acceleration load. Thead was specically a pressure of 1000mbar applied tof the diaphragm to simulate line pressure,with an addi-ure of 16mbar applied to the at side of the diaphragmthe small differential pressure being sensed. For thediaphragm with a hollow boss the inside of the bossed as being at 0 pressure. This simulates the vacuumoss that is expected due to the fusion bonding pro-

whichFig. 13

4. Cha

Toenoughlow-prsensititerizedsourcedie onsolelymadethe 25stress.inuengood inusing h

Diea fullyoffseto create it. The results shown in Table 1 demonstrateucing hollow bosses results in the highest sensitivityst linearity as compared to the alternatives. Sensitiv-ximately a factor of three better than the alternatives,r non-linearity to the solid boss diaphragm. The resultstion sensitivity indicate that while the at diaphragmsensitive as expected, the hollow boss device is onlymore sensitive and is a factor of 10 less sensitive thanss device. With these performance advantages in minds design was chosen to demonstrate the feasibility ofboss process. A micrograph of the fabricated sensor,

at diaphragm versus a diaphragm with hollow bosses.

Sensitivity(mV/V/mbar)

Linearity(%f.s.TBNL)

G-sensitivity(ppm of f.s./g)

80m height 1.16 0.48 110m height 0.37 0.50 114diaphragm 0.39 4 64 3

sensor full-units, as thpressure tomated to bconsistentwere mounand total drthe chart inable and insensors.

To test foa die was bential pressdevice usinsitivity of tFig. 15), wfull-scale (steresis on rthe good stface.position of the piezoresistive elements shown relative to the bosses,the bridge.

ased on the dimensions detailed above, is shown in

erization of hollow boss sensor performance

determine whether the hollow structures were stableuse in aMEMS sensor the performance of the fabricatedrediewasassessed. Thermal stability at 125 C, pressureand linearity, and pressure hysteresis were all charac-g the fabricated die. Packaging stress is a well-knownnsor instability, therefore to assess the stability of theecial care was taken to remove these affects. Die werehed to test electronics bymeans of the gold wire-bondse chip such that the die were essentially oating onold wires to minimise the affects of packaging inducedeasured stability would therefore be due solely to the

f the silicon die with hollow bosses and therefore be ator of the ultimate performance that may be achievedw bosses.e heated to 125 C in an oven, after approximately 8hle temperature was reached. The variation in 0mbarthis point was measured in percentage change of the

scale. The full-scale output was not measured for thesee oating construction did not allow for a differentialbe applied to the diaphragm. Therefore this was esti-e 18mV/V, which is a value of full-scale output that iswith the measured sensitivity for other die. Two dieted in this way and held at 125 C for a 450-h periodift was approximately 0.02% of full-scale, as shown byFig. 14. This level of drift was deemed to be accept-line with the performance of other MEMS pressure

r pressure sensitivity, linearity, andpressure hysteresisonded to a stainless steel packaged such that a differ-ure ranging from 0 to 30mbar could be applied to theg a low-pressure Ruska pressure controller. The sen-he device was determined to be 0.5mV/V/mbar (seeith a terminal base non-linearity of less that 0.4% ofee Fig. 16). There was no indication of pressure hys-eturn to 0mbar after pressure cycling, demonstratingructural nature of the hollow boss to diaphragm inter-

40 P.K. Kinnell et al. / Sensors and Actuators A 160 (2010) 3541

Fig. 12. Quarter sections of the three hollow bosses die, a single solid boss die, anda simple at diaphragm die.

Fig. 13. Micrograph of the hollow bosses shown on a silicon diaphragm.

Fig. 14. Ze

F

Fig. 16. Plot o

5. Conclus

A novelmounted tostructures hfor use inof the procof a micro-ro offset drift for two sensors over a 450h period spent at 125 C.

ig. 15. Plot of device output, with sensitivity indicated.

f terminal based non-linearity calculated for two hollow boss devices.

ion

process for the fabrication of hollow boss structuresthin silicon diaphragms has been developed. Theseave been employed to create hollow stiffening bossesa 30mbar full-scale pressure sensor. The feasibilityess to create structural elements that may form partelectro-mechanical systems (MEMS) device has been

P.K. Kinnell et al. / Sensors and Actuators A 160 (2010) 3541 41

demonstrated through the fabrication and testing a hollow teststructure that spanned a width of 400m with a height of 80mand a wall thickness of only 3m (see Fig. 7). This type of struc-ture may be used in a variety of MEMS applications for exampleas thin-walled encapsulations layers, thermal isolation structures,or as in this example for producing hollow stiffening structureson silicon diaphragms. The fabricated pressure sensor showed thesuitability of these structures for use as structural elements inMEMS sensors. This was demonstrated by the excellent thermalstability and absence of any pressure hysteresis that would not bepossible if the hollow structure was not a fully integral part of thesensor.

References

[1] W.C. Young, R.G. Budynas, Roarks Formulas for Stress and Strain, 7th ed.,McGraw-Hill, 2002.

[2] S. Franssila, Introduction to Micro Fabrication, Wiley, 2004, pp. 205216.[3] M.J. Madou, Fundamentals of Microfabrication the Science of Miniaturization,

2nd ed., CRC Press, 2002.[4] J.C. Greenwood, Etched silicon vibrating sensor, J. Phys. E: Sci. Instrum. 17 (1984)

650652.[5] M. Hirata, K. Suzuki, H. Tanigawa, Silicon diaphragm pressure sensor fabricated

by anodic oxidation etch-stop, Sens. Actuators A: Phys. 13 (1) (1988) 6370.[6] Y. Kanda, A. Yasukawa, Optimumdesign considerations for silicon piezoresistive

pressure sensors, Sens. Actuators A: Phys. 62 (1997) 539542.[7] J. Gardner, V. Varadan,O.Awadelkarim,Microsensors,MEMS, andSmartDevices,

John Wiley & Sons, 2001, pp. 126134.

Biographies

Peter Kinnell Peter Kinnell gained aMEng inMechanical Engineering from the Uni-versity of Birmingham (UK), and The Danish Technical University in 2001. Aftergraduating he continued his studies at the University of Birmingham undertaking aPhD in MEMS sensor design, specializing in advanced packaging for resonant straingauges. Since completing his PhD he has worked as a senior design engineer at GESensing. His work has included a range of products from large volume medical andautomotive sensor applications to high performance resonant pressure sensors.

Russell CraddockRussell CraddockgraduatedwithaBSc inChemistrybeforeunder-taking an MSc in Semiconductor Devices. He joined the Lucas Research Centre toinvestigate silicon pressure sensor and accelerometer design before transferring toLucas NovaSensor in theUSA towork on the development of automotive accelerom-eters. In 1992 Russell joinedDruck Ltd. nowGE Sensing, leading piezoresistive andresonant pressure sensors development for GE Druck and automotive pressure andaccelerometer products for GE NovaSensor.

Jim King Jim King graduated with a BSc (Open) with a Physics Diploma fromthe Open University. His semiconductor engineering background includes work-ing for Agilent Technologies fabricating lasers, photodetectors and diodes, CorningResearch developing semiconductor optical ampliers and electro-absorptionmod-ulators, Plastic Logic working on the development of polymer transistor arrays andexible displays. He has worked for GE Sensing since 2005 as a Senior Process Engi-neer developing and aiding manufacture of piezoresistive and resonant pressuresensors.

Mandy Lester Mandy Lester is currently studying for an MPhys degree at theUniversity of Manchester, where her research interests have ranged from ozonemeasurement by Brewer spectrophotometry to radio thin layer chromatographyfor medical PET imaging. She is due to graduate July 2010 and plans to pursue acareer in nuclear or renewable energy.

A hollow stiffening structure for low-pressure sensorsIntroductionThe hollow boss processTest sensor designCharacterization of hollow boss sensor performanceConclusionReferencesBiographies