JOURNAL OF MATERIALS SCIENCE35 (2000 )3447– 3457

Preparation and characterization of platinum

black electrodes

B. ILIC∗, D. CZAPLEWSKI∗Microfabrication Application Laboratories, University of Illinois at Chicago, USAE-mail: bi22@cornell.edu

P. NEUZILInstitute of Microelectronics, 11 Science Park Road, Singapore, 117685

T. STANCZYK‡Microfabrication Application Laboratories, University of Illinois at Chicago, USA

J. BLOUGHDepartment of Civil and Materials Engineering, University of Illinois at Chicago, USA

G. J. MACLAYMicrofabrication Application Laboratories, University of Illinois at Chicago, USA

We have investigated properties of electrochemically deposited platinum black by atomicforce and scanning electron microscopy. Platinum black was deposited on evaporatedplatinum electrodes. Deposition time and cure temperature was found to influence thequality and morphology of the platinum black layer. Morphological inclusions were readilyobserved in films deposited for duration of less than 60 seconds at a bias of 1.5 V against aplatinum counter electrode. Shorting of the microfabricated electrodes due to lateraloutgrowth of high surface area platinum black was observed when current densities on theorder of 100 mA cm−2 were employed. We further show that reproducibility of highlyadherent platinized electrodes is achieved. C© 2000 Kluwer Academic Publishers

1. IntroductionPlatinized electrodes, wherein a high surface area plati-num black is formed, have attracted attention in manyareas of chemistry, biology and physics [1–7]. Partic-ularly, high surface area noble metal electrodes suchas platinized platinum electrodes are used for a largevariety of microfabricated solid state, chemical and bi-ological sensors [8–21]. For example, a typical elec-trochemical sensor is composed of three electrodes: asensing or working electrode at which the reaction ofinterest occurs, a reference electrode that keeps the elec-trical potential of the working electrode constant duringthe measurement; and an auxiliary or counter electrodefor current injection in the electrolyte. In controlledpotential experiments, however, a single electrode canserve as both a counter and a reference electrode. Thiselectrode has a double function of passing current andcontrolling the potential of the working electrode. Ptis generally considered as both an inert metal whichdoes not enter an electrochemical reaction and a cat-alytic metal that will provide the proper kinetics whichincrease the rate of chemical reactions. In contrast toAu and Ag, Pt has the advantage that under positivepotential bias it is less reactive to Cl− ions in an elec-

∗ Present Address: Department of Applied and Engineering Physics, Cornell University, 212 Clark Hall, Ithaca, NY, USA.‡ Present Address: Molex Incorporated, 2222 Wellington Court, Lisle, IL, USA.

trolyte. Under these conditions Au and Ag electrodeschlorinate. Therefore, Pt electrodes are most commonlyutilized in chemical sensing applications.

The primary benefit of platinized electrodes is the in-crease in the surface area of the catalyst [22–25]. In thecase of the microfabricated CO sensor [20], the highsurface area Pt black consequently provides more sitesfor the oxidation of CO to CO2. As a result, the sensi-tivity is greatly enhanced. For instance, in contrast toplanar platinum and gold electrodes, enhancement toCO sensitivity by a few orders of magnitude was ob-served using a platinized Pt working electrode. Pt blackis also considered as one of the best materials for theoxidation of H2O2. It causes a reduction in the H2O2oxidation potential, which consequently increases theoperating stability of the biosensor and decreases inter-ference currents [26]. Platinized Pt electrodes have alsoplayed a role as a biosensor transducer and a matrix forenzyme immobilization [27].

The goal of this paper is to investigate effectsof electrodeposition conditions and annealing on thetopography and adhesion of platinized Pt electrodes.Topography is evaluated using tapping mode atomicforce microscopy (TMAFM) and scanning electron

0022–2461 C© 2000 Kluwer Academic Publishers 3447

microscopy (SEM). Methods of fractal analysis wereemployed to demonstrate the degree of roughness ofthe platinization.

2. Fabrication2.1. Substrate preparationThe silicon substrates consisted of three inch,〈100〉,p-doped prime grade wafers with 6–15Ä cm resistiv-ity. Prior to any processing, the wafers were cleanedin Summa Clean at 40◦C under ultrasonic agitation for20 minutes. Approximately 1 micron of silicon dioxidewas thermally grown in wet ambient (pyrogenic steam)at 1100◦C for 140 minutes. This layer served as an in-sulation layer between the doped Si wafer and the sub-sequent metal layer electrodes. Silicon nitride was alsoused as an insulating layer. Next, two hundred and fiftynanometers of silicon nitride (Si3N4) was deposited in alow pressure chemical vapor deposition (LPCVD) sys-tem using a mixture of dichlorosilane (SiH2Cl2) andammonia (NH3) at a temperature and pressure of 800◦Cand 250 mTorr respectively.

The Silicon wafers with oxide and nitride along withthe alumina substrates were cleaned in acetone for thirtyminutes under ultrasonic agitation. This was done pri-marily to remove any organic contaminants, which gen-erally hinder adhesion of Platinum. They were thencleaned in isopropyl alcohol for thirty minutes underultrasonic agitation to remove any residual organic con-taminants and acetone residue. A thirty-minute deion-ized water cascade rinse followed to remove any solventtraces. The wafers were further cleaned in a piranha so-lution (3 : 1 98% H2SO4 : 30% H2O2) for a duration of20 minutes. Again, the wafers were rinsed in a deion-

Figure 1 Process flow of the electrode fabrication and platinization. (a) Deposition of 1000 A of platinum via RF sputtering, (b) photolithographicelectrode definition, (c) platinum etching using aqua regia, (d) platinization of the platinum working electrode.

ized water cascade rinse to remove any acid traces andionic species. Substrates were nitrogen dried and placedinto an oven for 1 hour at 150◦C in order to dehydratethe substrates, thereby improving adhesion of Pt filmsto the substrates.

2.2. Platinum film depositionFollowing the clean, 1500̊A of Pt was deposited onthe substrates by RF sputtering (Fig. 1a). Initially, thesubstrates were placed into a CVC SC-4000 sputteringvacuum chamber. At a base pressure of<2× 10−7 Torr,argon (Ar) was introduced to both raise the pressure to5.5 mTorr and create a RF Ar plasma for sputteringof Pt. A five-minute predeposition at 110 W forwardpower was carried out in order to stabilize the plasmaand clean the Pt target. To achieve better uniformity,the wafers were rotated during the deposition process.

Wafers were then annealed at 650◦C for 1 hour in afurnace under nitrogen ambient. Annealing eliminateduse of metallic adhesion promoters, which can poten-tially perturb the electrochemical cell. Generally, ad-hesion promoters such as chromium or titanium havebeen found to degrade the sensing performance. Mostimportantly, annealing was found to be of indispensablesignificance to the adhesion of Pt to the substrates [28].

2.3. PhotolithographyPhotolithography was carried out to define the elec-trode geometry. First, Hexamethyldisilzane (HMDS)was spun at 4000 rpm for 10 seconds to assist photore-sist adhesion. Immediately following this spin, Shipley1818 positive photoresist was spun at 4000 rpm for30 seconds. The wafers were prebaked on a hotplate at

3448

Figure 2 Scanned image of the platinized Pt sensor. The black area overthe active portion of the electrode is Pt black while the gray is Pt.

90◦C for 5 minutes. Using a Karl Suss contact aligner,the photoresist was exposed to ultra violet light of amercury lamp through a photomask for 12 seconds atan intensity of 274 W/cm2 (Fig. 1b). Next, the pho-toresist was developed in a 1 : 4 solution of ShipleyMicroposit 351 developer and DI water respectively.

Figure 3 Current versus time characteristics during platinization of5 sensors.

Figure 4 Scanning electron micrographs showing peeling of platinized electrodes on alumina substrates.

To ensure resist durability during the Pt etch, the resistwas post-baked at 150◦C for 60 minutes. Platinum wasthen etched in a solution of boiling aqua regia for 4 min-utes (Fig. 1c). The solution consisted of 550 ml 37%HCl and 30 ml 70% HNO3 [29]. Finally, photoresistwas removed in a piranha solution, mixed to the speci-fications previously defined. This consequently cleanedthe Pt electrodes and the substrate of all contaminants.The substrates were then placed in a deionized waterrinse for 30 minutes to remove all acid components.

Following electrode definition, Shipley 1818 wasspun on the substrate at 4000 rpm for 30 seconds. Fol-lowing spinning, the wafers were cured on a hot plate at90◦C for 5 minutes. After curing, they were cut 4 mmwide by 11 mm tall using a dicing saw. The photoresistprotected the electrodes from damage during cutting.Following dicing, individual sensors were cleaned in apiranha solution. They were then rinsed in deionizedwater.

2.4. Platinum black coatingThe platenizing solution consisted of 3.5% chloro-platenic acid (Engelhard, Iselin, NJ) and 0.005% leadacetate. The working portion of the sensing electrodewas immersed into the platenizing solution in order toelectrochemically deposit high surface area platinumblack. Against a Pt counter electrode, a constant poten-tial of 1.4 V was applied for 190 seconds (Fig. 1d). Thissetup worked for single sensor deposition. However,a galvanostatic technique was utilized for batch pro-cesses, wherein platinum black is selectively depositedupon a strip of sensors or a whole wafer. Here, a con-stant current source output 30 mA/cm2 for 140 seconds.After deposition, Platinum Black was cured at varioustemperatures in nitrogen ambient. Fig. 2 shows a sensorwith platinized electrode.

Fig. 3 shows the current versus time characteristicsduring the deposition for several sensors. The area un-der the deposition curve yields the total charge con-sumed in the electrodeposition process which is pro-portional to the amount of platinum black deposited onthe sensing electrode. Assuming the same deposition

3449

efficiency during each deposition, through numericalintegration, variations in the total Pt black depositedwere less than 10%. The slight deviation of the platinumblack coating was primarily due to the uncontrolledvariation of the exposed area due to the immersion depthof the electrode, during electrodeposition.

(a) (b)

(c) (d)

Figure 5 Tapping mode atomic force topographs of various substrates. (a) 500 nm× 500 nm AFM scan of a clean Silicon (100) surface,(b) 500 nm× 500 nm AFM scan of LPCVD silicon nitride, (c) 500 nm× 500 nm AFM scan of an oxidized(100) Si wafer, (d) 15µm× 15 µmAFM scan of alumina.

Figure 6 5µm× 5µm tapping mode AFM image of sputtered platinum.

3. Results and discussion3.1. Adhesion of platinumAdhesion of Pt on the substrates is of critical impor-tance for further platinization and sensor fabrication.The adhesion of Pt was found to vary greatly with sub-strate preparation techniques and different substrates.

3450

Five types of substrates were used: silicon; thin film sto-ichiometric silicon nitride, low-stress silicon nitride andsilicon dioxide on silicon substrates; and alumina. Stoi-chiometric silicon nitride has a formula of Si3N4, whilelow stress silicon nitride has a higher silicon content, asseen in its formula SiNx wherex< 4/3 [30–32]. Thebiggest difference between the two materials is theirchemistry; however, the silicon industry utilizes thetwo materials because they exhibit differences in stress.The stoichiometric silicon nitride is known to have amuch higher stress than the silicon-rich low stress sili-con nitride.

The most critical step to ensure good platinum ad-hesion is the post-sputter cure at 600◦C. If the annealwas not carried out, regardless of the substrates or their

(a)

(b)

Figure 7 Tapping mode atomic force topographs of annealed platinized Pt electrodes. (a) uncured (5µm× 5 µm scan), (b) uncured (1µm× 1 µmscan), (c) 100◦C anneal (5µm× 5µm scan), (d) 300◦C anneal (5µm× 5µm scan), (e) 650◦C (5µm× 5µm scan). (Continued)

preparation, Pt would peel during either electrode def-inition or platinization. To determine initial film ad-hesion, the standard Scotch® Tape test was performedbefore any other processing step to rule out other causesof film de-adhesion [33, 34]. It was found that the testwould remove Pt from the surface of the wafers, therebyshowing poor adhesion following sputtering.

The second critical step to prevent peeling of thePt film is a pre-sputter dehydration of the substrates.Under air ambient conditions most surfaces will ac-cumulate several monolayers of adsorbed gases suchas condensed water vapors, that is why the substrateswere placed in a 150◦C oven for dehydration followingthe pre-deposition cleaning. If this step was not carriedout, again, the electrodes would peel during electrode

3451

(c)

(d)

(e)

Figure 7 (Continued).

3452

Figure 8 Fractal dimension versus Pt black anneal temperature plot.

(a)

(b)

Figure 9 Scanning electron micrograph of non-uniform lateral outgrowth of platinum black at various magnifications: (a) 2000×, (b) 510× of a10µm spacing interdigitated electrode configuration, (c) 1400× of a corner. (Continued)

definition or platinization. An example of peeling fol-lowing platinization is shown in Fig. 4a and b.

Generally, to enhance adhesion, it is highly advanta-geous to include a layer of strong oxide forming elementbetween the substrate and metal. Intermediate adhesionlayers such as titanium and chromium are typically usedfor this purpose. However, it is critical that adhesionlayers is not used to aid in the platinum adhesion sincelayers such as titanium and chromium cause corrosioncurrents in most applications, which prevent accuratemeasurements of potentials or currents.

Platinum adhesion also depends on substrate choice.Following adequate surface preparation, cleaning, rins-ing, and dehydration, it was observed that platinum didnot adhere well to silicon oxide and stoichiometric sili-con nitride, because local reactions of the materials didnot take place. However, on Si, low-stress silicon nitride

3453

(c)

Figure 9 (Continued).

and alumina, properly prepared substrates did not ex-hibit peeling or other signs of poor adhesion. In the caseof Si, it is common that a platinum-silicon compoundforms at 600◦C, where platinum reacts with silicon toform a platinum silicide (PtSi) at the interface. This isalso true for low stress silicon nitride since the layeris silicon rich. However, on the alumina substrates, ex-cellent adhesion was observed due to a high surfaceroughness. This facilitates mechanical interlocking totake place between Pt and Al2O3. Surface roughnessdid not figure into the adhesion on the silicon, siliconnitride and silicon dioxide surfaces since they are ex-tremely smooth compared with the surface of alumina.From AFM imaging, the RMS roughness for Si, SiO2,Si3N4 were respectively in the range 0.6–0.8 A, 4–6 A,and 8–16 A. Fig. 5a–d shows atomic force micrographsof the surfaces of the substrates, where it is apparent thatsurface roughness is large for alumina and small for theother substrates.

3.2. Platinized electrodes analysis3.2.1. Grain sizePlatinum black electrodes were evaluated using bothatomic force and scanning electron microscopy. An-neals at various temperatures were carried out and theaverage grain area was determined using an AFM in tap-ping mode with oxide sharpened silicon tips [35, 36].Ordinary profilometry and contact mode AFM imag-ing could not be used to image Pt black since at thelowest force setting it was strong enough to damagethe surface. Fig. 6 shows an AFM image of a smoothplatinum surface prior deposition while Fig. 7a–e showAFM images of platinum black films cured at severaltemperatures. Tapping mode AFM images show a ran-domly scattered nanogranular structure of which the

cauliflower platinized structure is composed. In partic-ular, Fig. 7b shows an 1× 1 µm AFM topograph ofan uncured platinized film. Grains on the order of tensof nanometers are seen. These nano-sized grains coa-lesce to form larger grains at elevated cure temperaturesand almost completely disappear when the curing tem-perature is above 650◦C (Fig. 7e). Therefore, from theAFM images, it is apparent that the higher the cure tem-perature the larger the resulting grain size. This furtherimplies that at higher cure temperatures the surface areais lowered.

3.2.2. Fractal imagingFractal dimension analysis can be used to mathemati-cally characterize the geometric complexity of the sur-face. The fractal dimension of a three dimensional sur-face varies from a minimum of 2.00 for a flat surface to amaximum of 3.00 for an infinitely rough surface. DigitalInstruments software utilizes an algorithm which com-prises of a triangular form of analysis [37]. Fig. 8 showsa plot of the fractal dimension versus the Pt black curingtemperature. The dependence of fractal dimension de-creases with increasing Pt black cure temperature whichalternatively signifies a decrease in both roughness aswell as surface area with increasing cure temperature.This is because higher anneal temperatures result in anincreased surface mobility which leads to filling of thepeaks and the valleys, thereby decreasing the surfaceroughness.

3.3. Deposition time during platinizationDeposition time, during platinization, was found tobe important factor for the uniformity and adhe-sion of platinum black. From Fig. 9a–c, non-uniform

3454

Figure 10 Scanning electron micrograph at 1450× of voids in Pt black.

outgrowth of platinum black for several geometries wasshown to be on the order of a few microns for deposi-tion current densities greater than 100 mA cm−2. Also,even at low fields, shorting of the 20µm gap electrodeswas readily observed for deposition times exceeding10 minutes. This is primarily due to a large electricfield concentration at the edge of the planar electrode.From, the Appendix A derivation of the field at the cor-ner of our electrodes, Equations A.4a and A.4b showelectric field line crowding at the edge which is syn-onymous with a local increase in the electric field thatis responsible for lateral outgrowth.

Voids, as shown in Fig. 10, on the order of severalmicrometers were observed throughout the electrodewhen the deposition time was less than 60 seconds. Thesize and distribution of the voids were random. Thisis a consequence of a nonhomogeneous nature of theplatinum black deposition process. Since the observedlateral encroachment is on the order of the void-size,given enough time, lateral encroachment will eventu-ally close the gap.

4. ConclusionsAtomic force and scanning electron microscopy wereused to investigate the topography of annealed micro-fabricated platinized Pt electrodes. Surface propertiesof the electrochemically prepared high surface area Ptblack films were found to have a strong dependence on

electrode preparation prior electrodeposition. Scotchtape tests showed stoichiometric silicon nitride and sil-icon dioxide to exhibit poor adhesion. On the otherhand, Si, low stress silicon nitride and alumina sub-strates showed excellent adhesion. Si, SiO2 and Si3N4substrates are relatively smooth and in order for ad-hesion to take place an intermetallic platinum silicideforms. It is known that oxide and stoichiometric nitridedo not form silicides with Pt. However, due to the highsilicon concentration in the low stress Si3N4, it is pos-sible that a Pt-silicide was formed and thus enhancedadhesion.

Initial surface roughness of the substrate was alsofound to affect the adhesion of the electrodes. Increasedsurface roughness may promote adhesion because thesubstrate exhibits more surface area than a flat surface,and mechanical interlocking between the substrate andfilm may occur as in the case of alumina. Adhesion alsodepends upon the cleanliness of the substrate. Contam-ination generally results in reduced adhesion, as doesan adsorbed gas layer. A dehydration treatment wasfound to be essential in order to remove any adsorbedwater vapors. Cleaning the substrate prior depositionwas important to assure film adhesion. Platinum blackelectrodes generally exhibit morphological anomaliesthat are alleviated by a piranha pretreatment.

Deposition time played an important role for the uni-formity and adhesion of Pt black. Non uniform out-growth of platinum black was shown to be on the order

3455

of a few microns for deposition current densities in ac-cess of 100 mA cm−2. The lateral outgrowth was dueto the electric field line crowding at the edge of theelectrodes. Therefore, we can see that excessive pro-trusions of platinum black can play a major role whenconsidering electrode spacing ranging from a few totens of micrometers. On the other extreme end of thespectrum, when the deposition time is sufficiently short(<60 seconds), large voids were readily seen through-out the electrode due to possible passivating contami-nants.

Finally, we have used AFM of the platinized workingelectrode toshow a direct cure temperature correlationwith respect to the average grain size. Larger grainswere observed for higher anneal temperatures.

Appendix A: Electric field crowding observedat cornersConsider an intersection of two conducting planes heldat a potentialV , as shown in Fig. A1, that define acorner in two dimensions with an opening angleα. Inthe polar coordinate system (ρ, φ), Laplace’s equationin two-dimensions is given by,

1

ρ

∂

∂ρ

(ρ∂8

∂ρ

)+ 1

ρ2

∂28

∂φ2= 0 (A.1)

where the general solution may be expressed as,

8(ρ, φ) = V +∞∑

l = 1

a1ρlπα sin

(lπφ

α

)(A.2)

For small enoughρ only the first term in the seriesbecomes important, therefore nearρ= 0 the potentialis approximately,

8(ρ, φ) ∼= V + a1ρπα sin

(πφ

α

)(A.3)

The electric field components are given by,

Eρ(ρ, φ)=−∂8∂ρ=−πa1

αρ

(πα

)−1 sin

(πφ

α

)(A.4a)

Figure A1 Intersection of two conducting planes held at a potentialV .

Figure A2 Special case of the opening angle producing an outsidecorner.

Eφ(ρ, φ) = − 1

ρ

∂8

∂φ= −πa1

αρ

(πα

)−1 cos

(πφ

α

)(A.4b)

The case ofα= 3π/2 is shown in Fig. A2. At this an-gle the corner becomes an edge, as in the case of ourelectrodes, where the field becomes singular asρ→ 0.

AcknowledgementsThe authors would like to acknowledge F. MichaelSerry (Digital Instruments, Santa Barbara CA) andTony Cocco (University of Illinois) for their helpfuldiscussions. The authors would further like to thankClark E. Smith, Ed Duran, and Mark Devine at FirstAlert for their support and Jiri Holoubek from IST AG,Switzerland for useful hints regarding the Pt depositionand treatment.

References1. W. M . M A C N E V I N andM . L E V I T S K Y ,Anal. Chem.24(1952)

973.2. W. L A N G , K . K U H L andH. S A N D M A I E R, Sens. Act. A34

(1992) 243.3. B . W R I G H T andE. B. P A T T I N S O N, J. Phys. D: Appl. Phys.

7 (1974) 1560.4. J. R. L O S E E andD. S. B U R C H, Rev. Sci. Inst.43 (1972) 146.5. J. J. R U S H, R. R. C A V A N A G H andR. D. K E L L E Y , J. Vac.

Sci. Technol. A1 (1982) 1245.6. F. C. A N S O N andD. M . K I N G , Anal. Chem.34 (1962) 362.7. F. C. A N S O N, ibid. 33 (1961) 934.8. J. J. P A N C R A Z I O, R. P. B E Y, A . L O L O E E, S. M A N N ,

H. C. C H A O, L . L . H O W A R D, W. M . G O S N E Y, D. A .B O R K H O L D E R, G. T. A . K O V A C S, P. M A N O S, D. S.C U T T I N O andD. A . S T E N G E R, Biosensors and Bioelectronics13 (1998) 971.

9. S. Y A M A U C H I , M . Y A O I T A , F. M A T S U M O T O, T.Y O K O Y A M A and Y . I K A R I Y A M A , Sens. Act. B13 (1993)79.

10. K . W. J O H N S O N, ibid. 5 (1990) 85.11. R. B. B E A R D, J. F. D ER O S A, R. M . K O E R N E R, S. E.

D U B I N andK . J. L E E, IEEE Trans. Biomed. Engin.19 (1972)233.

12. T . S U D A andS. U E N O, J. Appl. Phys.85 (1999) 5711.

3456

13. D. R. J U N G, D. S. C U T T I N O, J. J. P A N C R A Z I O,P. M A N O S, T. C L U S T E R, R. S. S A T H A N O O R I, L . E.A L O I , M . G. C O U L O M B E, M . A . C Z A R N A S K I , D. A .B O R K H O L D E R, G. T. A . K O V A C S, P. B E Y, D. A .S T E N G E R and J. J. H I C K M A N , J. Vac. Sci. Technol. A16(1998) 1183.

14. R. J. R E A Y, R. D A D O O, C. W. S T O R M E N T, R. N. Z A R E

andG. T. A . K O V A C S, in 1994 Solid State Sensors and Actu-ators Workshop, June 13–16, Hilton Head, South Carolina, 1994.

15. S. R. S A M M S, S. W A S M U S andR. F. S A V I N E L L , J. Elec-trochem. Soc.143(1996) 1498.

16. J. K U W A N O , A . W A K A G I andM . K A T O , ibid. 139 (1992)L113.

17. T . O Y A B U , T. O S A W A andT. K U R O B E, J. Appl. Phys.53(1982) 7125.

18. R. V . K U M A R andD. J. F R A Y, Sens. Act.15 (1988) 185.19. P. V A N Y S E K , Can. J. Chem.75 (1997) 1635.20. G. J. M A C L A Y , D. K E Y V A N I andS. B. L E E, in Proceed-

ings of the Second International Symposium on Microstructures andMicrofabricated Systems, Chicago, October 1995 (ElectrochemicalSociety Proceedings) Vol. 95-27, p. 177.

21. S. B. L E E, A . C O C C O, D. K E Y V A N N I and G. J.M A C L A Y , J. Electrochem. Soc.142(1995) 157.

22. T . B I E G L E R, ibid. 116(1969) 1131.23. M . J. J O N C I C HandN. H A C K E R M A N , ibid. 111(1964) 1286.

24. C. A . M A R R E S E, Anal. Chem.59 (1987) 217.25. A . M . F E L T H A M andM . S P I R O, Chem. Rev.71 (1971) 177.26. G. F. K H A N andW. W E R N E T, J. Electrochem. Soc.143(1996)

3336.27. Y . I K A R I Y A M A , S. Y A M A U C H I , T . Y U K I A S H I and H.

U S H I O D A, ibid. 136(1989) 702.28. J. H O L O U B E K, private communication.29. P. W. H. W A L K E R , T E R R I N (eds.), “Handbook of Metal

Etchants” (CRC Press, 1991).30. S. L . Z H A N G, J. T . W A N G andW. K A P L A N , Thin Solid

Films 213(1992) 182.31. T . M A K I N O , Solid State Sci. and Technol.130(1983) 450.32. S. H A B E R M E H L, J. Appl. Phys.83 (1998) 4672.33. P. A . S T E I N M A N N andH. E. H I N T E R M A N N , J. Vac. Sci.

Technol.A7 (1989) 2267.34. J. V A L L I , ibid. A4 (1986) 3007.35. R. W. S T A R K, T . D R O B E K andW. M . H E C K L , Appl. Phys.

Lett.74 (1999) 3296.36. A . K U H L E , A . H. S O R E N S E NandJ. B O H R, J. Appl. Phys.

81 (1997) 6562.37. Digital Instruments Nanoscope III Command Reference Manual.

Received 5 Octoberand accepted 14 December 1999

3457