Transcript

TUIA-4

EVALUATION OF TECHNIQUES FOR BONDING WIRES TO QUANTIZED HALLRESISTORS ·Dr Kevin C Lee

National Institute of Standards and TechnologyTechnology Administration

U.S. Department of CommerceGaithersburg, MD 20899

Abstract

Three different techniques for mountingquantized Hall resistors with AuGe/Ni alloyed contactswere evaluated. The best quality and most robustsamples were made by evaporating bonding pads thatoverlapped the alloyed contacts and the substrate. sothat bonds could be made over the substrate rather thanover the heterostructUre.

Introduction

Quantized Hall resistors made from GaAs/AlGaAsheterostructures with alloyed AuGe/Ni contacts havefoundwide applicationas resistancestandardsbecausethey are of high quality. and can be mass-produced.The close proximityof the electrongasresponsibleforthe quantum HaD effect to the surface of theheterostructure. however. makes it difficult to attachwires to the contacts without degrading them. Thisproblem is made more difficult by the extremes oftemperature and fairly high stresses that the samplesexperience.particularlywheninsertedintoa colddewarduring the cooling process. These conditionsrequirethat the wires be attached to the pads fmoly enoughthat gusts of helium gas evolved during the coolingprocess do not cause the wires to become detached. andthat the adhesive used to attach the sample to theheader remains adherent between room temper.nW'eandliquid helium temperature. In addition. quantized Hallresistance devices to be used as resistance standardsshould' be highly reliable. and be capable of being usedfor many years without degrading or requiring repairs.

Exnt!l"imental Technioues

In order to meet these challenging requirements. severaldifferent mounting techniques have been devised andevaluated. Wires have been attached to the sampleusing the following techniques:

1.1. Soldering1.2. Direct Wire Bonding1.3. Enlarged Bonding Pads

The samples have been attached to headers using thefollowing techniques:

· U.S. Government work not subject to U.s. Copyright.

t This worle supported in pan by the CalibrationCoordination Group of the Deparunent of Defence

2.1. Silicone vacuum grease2.2 Paraffin-impregnated plastic film2.3. Conductive epoxy

The quantized Hall resistors used in this study wereproduced by the Laboratoires d'Electronique Philips(LEP) under contract to the EUROMET Consortium[1]. These samples have alloyed AuGe/Ni contactswith a Ti/PtIAu thickening layer over the contacLExtensive tests on samples mounted using each of theabove techniques show that while all can be used toproduce standards-quality samples. the fust twotechniques (1.1 and 1.2) have disadvantages that makethem less desirable for mounting samples that are to beused as resistance standards for long periods of time;Samples made using the third technique (1.3) haveproven to be the r:nost reliable and of the highestquality. The advantages and disadvantages of each ofthe three techniques are discussed in the next section.

Summary nf Re!l;ults

Solderin!z: Gold wires with 2S ~ diameter weresoldered to the bonding pads using indium solder. Thewires attached using this technique are very fmolyattached to the pads. This technique does not requirethat the pads be extremely clean prior to soldering, andany of the three techniques (2.1-2.3 above) formounting the sample in the header can be used. Thisis the easiest technique to use. and gives reliable. lowresistance contacts. Unfortunately. however. indiumand gold readily fonn intermetallic compounds [2]which are quite brittle and readily fracture under thennalor mechanical stress. Experience at NIST with othersamples that have gold-indium solder connections indi-cates that while the solder connection may be quite.strong initially, these intermetaUic compounds fonnover periods of a decade or more. and the connectionseventually break. Thus. from the point of view oflong-term reliability, this technique is less desirable formounting resistance standards.

Direct Wire Bondin~: Gold wires with 2S ~diameter have been bonded to the gold pads directlyabove the heterostructure. This. however. is anextremely delicate task. The 2 dimensional electrongas responsible for the quantum Hall effect is onlyabout 60 nm below the surface. so any damage createdduring wire bonding will directly affect the propertiesof the device. In facL bonding pressures of as little as

112

73.5 MFa; created elecbically active defects in GaAs.even if there was no applied ultrasonic power [3].Since the defects act as acceptors. they raise theresistance of the contacts. and can cause them to ceaseto carry current at low temperatures. In order tosuccessfully bond wires to the pads. therefore. thelightest possible bonding forces must be used. Underthese conditions. however. the wire will not stick tothe sample unless both the wire and the bonding pads.are of the highest cleanliness.

In addition. the sample must be held fmnly duringbonding. as any movement can result in damage to thesample or even fracture of the GaAs. As a result. itwas found necessary to use epoxy to secure the sampleto the header (or to a glass or ceramic plate latermounted in the header) during bonding.

In spite of the extreme precautions taken during wirebonding. samples mounted using this techniqueshowed a degradation in room temperature contactresistance after bonding. The samples still proved tobe usable as resistance standards. The light pressuresused to attach the wires to the sample. however. didnot attach them sufficiently wen to withstand morethan one cool-down. This. together with the slightdegradation in the contacts on bonding. makes thistechnique the least desirable for mounting standards-quality samples.

Enlar!ZedBondinszPads: The technique that resultedin the highest quality samples was to evaporate goldbonding pads over the existing AuGelNi pads in such amanner that they overlapped the substrate outside ofthe Hall bar. as shown in Fig. 1. Wires could then bebonded to the pads over the substrate. so that anydamage produced during bonding would not affect theelectrical properties of the contact.

~:r:t:@1AuGe/Ni Contacts

~ Enlarged "BondingPads.

Fig. 1. Diagram of samples. The grey rectangles arethe original AuGeJNi contacts. and the cross-hatchedrectangles are the gold bonding pads. The enlargedpads overlapboth the AuGe/Ni contacts and thesubstrate. eliminating the need to perform the bondingoperation over the sensitive heterostructure.

; This pressure corresponds to a bonding force of lessthan 9 gm on a 25 j.1mdiameter wire.

" -

Samples bonded using this technique exhibited nochanges in room-temperature contact resistance onbonding. and showed the best contact resistances whenmeasured under quantum Hall effect conditions.Precision tests of these samples show that they are ofstandards-quality. The wires are very fiImly attached tothe sample. and can withstand the thermal andmechanical shocks that these samples are exposed to innonnal use. Current experience indicates that thesecontacts should not deteriorate with time. permittingthese samples to be used for long periods without theneed for repairing leads.

ConC!ID~inn

Soldering and wire bonding were evaluated astechniques for attaching wires to high quality quantizedHall resistors with AuGelNi alloyed contacts. Thesoldering technique was the easiest to implement. butsamples made using this technique potentially sufferfrom long-term degradation due to the formation ofintermetallic compounds at the gold-solder interface.The highest quality. most resilient samples wereprepared by depositing bonding pads that overlappedthe AuGe/Ni contacts and the substrate. enablingbonding to be performed over the substrate so thatdamage to the brittle GaAs during bonding did noteffect the electtical quality of the contact. Samplesprepared in this manner had the highest breakdowncurrents and lowest contact resistances of any of thesamples made using the other techniques.

Referenee!l

[1] F. Piquemal. G. Geneves. F. Delahaye, J.P. Andre, J.N.PaIillon. and P. FrijIink. "Repon on a joint BIPM-EUROMETProjectfor the Fabricationof QHEsamplesby the LEP". IEEE Trans. Instr. Meas..VoL42. No. 2.pp.264-268. April 1993.

[2] M. Hansen.Constitutionof Binarv Allovs.New York:McGrawHill Book Co.. pp. 240-1. 19S8.

[3] F. Hasegawaand H. Ira. "Degradationof a GunnDiOOeby DislocationsInduced During ThennocompressionBonding."AppLPbys. Lett.. VoL 21. No.3. pp. 107-8. August 1972.

113


Top Related