T, W, and / Are The Thickness, Width and Length of The Cantilever Beams, Respectively
T, W, and / Are The Thickness, Width and Length of The Cantilever Beams, Respectively
T, W, and / Are The Thickness, Width and Length of The Cantilever Beams, Respectively
inwhichEistheelasticmodulus,i.e., theYoung'smodulusofthebulkmaterial,and
t,w,and/arethethickness,widthandlengthofthecantileverbeams,respectively.
Thewidthandlengthofthecantileverbeamscanbedetermined quiteaccuratelyand
thevariationinthesevaluesforthesametypeofcantileversononewafer isvery
small.However,cantileversareoften verythinand,asstatedbefore,theremaybe
significant differences inthethicknessbetweencantileversononewafer. Sincein
equation2.2thethicknessiscubed,thismayleadtolargedeviationsfrom theactual
springconstantvalues.Todeterminethespringconstantwouldnowinvolve
measuringthethicknessoftheindividual cantilevers.Evenifthiswereasimpletask
thespringconstantmight stillbequestionablebecausetheapplicabilityofthe
calculatedvaluealsoreliesontheaccurateknowledgeoftheelasticmodulus.This
modulusvariesdueto(local)variationsinthestoichiometryofthesiliconnitrideand
thatofthegoldcoating,whichareunavoidable inmicrofabrication techniques.
Toavoidtheuseofa(toosimple)modelforthecantilever shape,SaderandWhite[9]
performed afinite elementanalysisofthestaticdeflection ofV-shapecantileversand
presented exactnumericalresultsfor thespringconstantfor avarietyofcantilever
dimensions.However,alsointhisapproachtheresultsdirectlydependonaccurate
valuesforthethicknessandelasticmodulusofthecantilever,whicharenoteasily
accessible.Thisrestrictstheapplicabilityofanymodelcalculationofthespring
constant.Inpractice,measurementofthespringconstantoftheindividual cantilever
isstillthebestoptiontogetquantitativelyreliableresultsfrom force measurements.
2.3.2 TheClevelandmethod
InthisworktheClevelandmethod[6]wasusedfor calibratingthecantilevers.Inthis
method,particlesofknownsizeanddensityareattachedtothefree endofthe
cantileverandthespringconstantkisdetermined from theresulting shift ofthe
cantilever'sresonancefrequency. Usingthismethod itisnotnecessarytoknowthe
massofthecantileveritself.
Theresonancefrequency v0 oftheunloadedcantilever isgivenby:
1 [k
VoV- (2-3)
2K \ m
28
ColloidalProbeAFM-Experimental Methodology
4P 2n\m +
M
(2.4)
Rearrangingthisequationgives:
M- (2.5)
(2nvf m
Thus,measuringtheresonancefrequencies ofthecantileverforvariousattached
massesandplottingMas afunction of (27tv)'2 shouldgivealinewithaslopeequalto
thespringconstant.Ifdesired,theeffective cantilevermassfollows from theintercept
ofthisplotfor 1/v2 = 0 .
Theresonancefrequency ofthecantilever canbeeasilydetermined makinguseofthe
NanoScope software. Firsttheresonancefrequency v0 oftheunloaded cantileveris
determined.After thisaparticleofknownmassisattachedtothecantilever,andthe
resonancefrequency isagainmeasured.Attachingtheendmassisdoneinmuchthe
samewayaswiththecolloidalprobes,onlythistimenoglueisused(inairthe
particlesstickduetocapillaryadhesion).Asendmassesweusedtungstenspheres
withdiametersintherange 10-30um(seefigure 2.4).Themassesofthespheres
werecalculated fromthesphereradiiandthedensityoftungsten(19.3103kg/m3).
Figure2.4: Scanningelectronmicrographof
thetungstenparticlesusedincalibratingthe
cantilevers.
29
Chapter2
Inprinciple,measuringtheresonancefrequencies oftheunloadedcantileverandthe
loadedcantileverforjust onevalueofMwould suffice tocalculatekfrom equations
2.3and2.4.However,inpracticemoremeasurements areneededtoreducethe effect
ofexperimental errors,inparticularthoserelatedtodeterminingtheprecisesizeofthe
tungstenparticles.Anexampleofaspringconstantcalibrationplotisgivenin figure
2.5.
1/(2TTV) 2 (10"1 S 2 )
2.3.3 Estimationofthespringconstant
Forcantileversdiffering inthicknessbutotherwiseidentical,thespringconstantsare
proportionaltothecubeoftheirunloadedresonancefrequencies. Thisisexpressedin
thefollowing equation[6]:
k=2ieiW2E-*\v0i (2.6)
30