A Selective Etching Solution for Use with Patterned Self-Assembled Monolayers of Alkanethiolates on Gold

2332 Reprintecl from Chemistrv of NIA'I'ERIALS.1995. 7.

Younan Xia, Xiao-NletZhao, Enoch Kim, and GeorgeNI. Whitesides* Departmentot''Chemislrl', Haruard (Jnit,ersity,Camhridge, Massachusetts 02138

Receit,edMay 2,3,1995. ReuisedMonuscript Receit;edSeptember 25, 1995!

This paper describes a selective etching solution fbr use with patterned self-assembied monolayers (SAMs) of alkanethiolates on the surface of gold; it is also effective for SAMs on the surfaces of silver and copper. This etching solution use'sthiosulfate as the ligand that coordinates to the metal ions and ferricyanide as the oxidant. This etchant has a number of advantages relative to the system of cyanide ion in oxygen-saturated, alkaline water used in previous work. First. it is less toxic, Iess hazardous. and has smaller environmental impact. Second,using it. complete etching of bare gold can be achieved more rapidly (-B min versus -15 min for 200 A of Au). Third, etching of bare gold occur-sr"'ith fewer def'ects in the SAM-covered regions and generates features of gold w,ith higher edge resolution. The influence of the composition of the etching solution on the rate of etching was studied svstematicallv.

Introduction The toxicitv and environmental impact of an etchant that contain-.free c1'anideion are obviouslv unattrac- Nlicrocontactprinting trrCPtr is a technique that tive. IIere,lve. report a more selectiveetching solution generatespatterned self-assembledmonolayers lSAtr{s t2 for use with SA\,{sof alkanethiolateson gcildthat uses of'alkanethiolates ttypically, hexadecanethiolate ) on the thiosulfate rather than c;ranide as the coordinating surfacesof goid,1silver,ll and copper,land of alkylsilox- ligand. and ferricyanide rather than O: as the oxidant. anes on hydroxyl-terminated surf'aces.r'Patterned SAI{s Ferrilferrtrcyanideare far less toxic than free cyanide- of alkanethiolates serve as nanomc.ter-thickresists in values of' LD;ri (ORL-MUS) are KC^ (8.5 mglkg),14 certain etching solutions: the one we i.rave.used most K,tFe(ClJ),:(2970 mg lhg),ta and K$e(CN),; t5000 mg I extensivelywhen gold is the underiving substrate has k91.":' beenaqueous KCN solution(pll'. 1-t.ICN ] ..,0.1M) This f'ern/f'errocvanideetchant is also superior to that saturated with 0z.6 Llsing this etchant. tl-repatterns containing CIN in rts selectivity: it is more rapid. in the SAMs can be transf'erredfaithfuih' into the thin generatesf'eq'er defects in the SAM-coveredregions and films of gold; the resulting patterns of gold can be used yieids microstructuresof gold with higher edgeresolu- subsequentlyas secondarvmasks for the etching (iso- tion. tropic or anisotropic) of' underlving la.versof silicon dioxide and silicon;' or ils functional elements (for Experimental Section example,for microerlectrodeseand diffractiongratingsi0). printing Materials. Au 199.999?),Ag t 99.9999q(),Cu(99.9999( ),Cr- Microcontact is renrarkablvconvenient fbr the (>99.99'i ,. - t.Tit99.99(i K:S:O'i, KSCN, KrFe(CN)c, KrFe(CN)0, fabricationof structuresrvith featuresizes of 1 zrmand andCII'rrCli, r, ,SH 'nvc.re obtained from Aldrich. KI. KOH.and larger: it can be r-rsed.rvith more difficulty and less K('N n'r're purt'liirsedfrom Fisher Scientific. Poll'(dimethyl- reliabilitv. to fabricatef'eatures with sizesdown to - 100 siloxane',rl'DlISt u'as obtainedfrom Dow Corning iS;zlgard nm.'"1tt l.r 1.84).Polished Sirl00trvafers (Cz, N/phosphorous-doped. 1- 10 Q cm. test gnrde.SEMI Std. flats)were obtainedfrom Silicon Sense. Inr'. ,Nashua. NH). I{exadecanetLiiolwas purified s Abstract publisheclin Arft'cuiccACS A6.s1rccls.Novcmber 1. 199ir. undcr nitrogr:n bv chromatographythrough silica gel. Gold (1) See.{or t,xample:Wilbur. .I. I-.:Kumar. A.: Kirn. ll.: \\hitcsides. iilms t200 ,\ thick) were preparedusing thermal evaporation G. N'1.Arft XIattr. f 994. 6. 600. (Varian 3118) onto silicon wafers rn'hosesurfaces had been (2) Rervicrvof'SANIs: \lhitesides, (i. NI.:Laibinis, E. E. Lortgtrtttir (-50 Gold. f 990, 6. 87. Dubois.L. H.; Nuzzo.R. G. Artrtu.Rer'. /)it.r'q. ('ltent. 1992. primed w'ith thin la.vc'r-sof chromrum A thickr. 1,:).4il7 iilt'". and copperfilm-q r500-1200 A thickr were prepared t3 t Xia, Y.; Kim. E.: \\rlritesides.(i. \I. ,J.ElectroL ltcni. Sor'.. in prerss. usirrge'-beam sputtering onto tita'rium-primed (- 15A ) silicon ('1) Xia. Y.: Kim. Ii.: \Irksich. !1.; \{tutesides.G. NL ('lrr'ni.tr7oter.. wafers. We chose200 A thick gold films for most of the current Submitted. work for three r€'Asons:(i)Gold films '200 A seemto behave r5r Xia. Y.; NIrksich.NL; Kim, E.: Whitesides.(i. N{.J. Ant. Chent. verS'similarh' in ferri/ferrocyanide etching solutions; thinner Sor'.1995, I17,9516. (6t Kumar. A.; Biebuyck, H.; Abbott. N. L.; \\hitesidc's.(;. NI. ./. fihns have more defects: 200 A is thereforea "standard thin Arn. Chent.Soc. 1992. 11J.9188. film".r4 tii) Gold is relativelyexpensive, and we wishedto use t7) Kendall,D. L. Attnu. Rt'r.Moter. Scr. 1979,9.:17:1. I)cterson. K. E. Proc. IEEE 1982.70.'120. {8) I{rm. E.: Kurnirr. A.:\Vhite'side's.(i. \'I. .1.ELectrocltern. Sor' 1995. i13r Wilbur. .I. L.: Kim. E.: Xia. Y.: Whilesides.G. NI.Adrr. Mater. 112,628. 1995. 7. 6.19. t9t Abbott,N. i,.: Rolison.I). R.;Whitesides, G. Nl.LctrLgnutir 1994. illt Mott'riol SrLfet.tDuta Sheets (CD-ROMr. Sigma-Aldrich- 8.2672. t'luka: rv'lilw;ruki.e.WI. 1994. (10) Kumar. A.; Riebur-ck.H.; \\tritcsidcs, (1. \f . Lortgmuir1994, (15t Moterial Safett'I)nta Sheets(CD-ROMr. NIDL Infbrmation 10. 1498. Svstcms.Inc.: WashingtonDC, 199,1. (11) Xia. Y.; \trhitcsides.G. NI..1. Ant. Chent.Stx. 1995./ 17.327,+. t16t Zhao. X.-NL; Wilbur, J. L.: \\hitesides, G. X,I.,unpublished (12) Xia. Y.: \\rhitesides.G. NI. Adu.Mater. 1995.7.,+7t. rcsults.

0897-4 756 I 9512807-2:13 2S09. 00/0 O 1995 American Chernical Socictv Monol,aversof Alkanethiolates on Gold

Au/SAM No SAM 1""(arb. unit; + t (tinescan) €.3 a 3min c 3.0 r :t \- z.z ri2 I (UL. I o - o1 ol ? b 7 min 0.0 5.0 10.0 15.0 20.0 2s.0 3.0 etching time (min)

0.6

'Etr .E c 10 E .r.N iH 5

0.0 0.05 0.1 0.1s lK2s2o3l(M) r- 25

lum 20 I Figure 1. SEMs of gold films (200 A thicki that had been Frs patterned with SAMs of hexadecanethiolate by rrCP, and E etchedin the standard etching solutioncontaining SzO,r2 and --E or ferrilferrocyanide for different intervals of time. Note that the 10 density of pits in the SAM-coveredregions increased as etching time increased. The intensity of the secondary electrons (1.", shown schematically as a line scan in arbitrary units) can be used to monitor the etching processsemiquantitatively. The scansdrawn on the right side were along the lines marked in 0.5 1 1.5 the SEMs on the left side. tKoHl(M) 10 c'oldfilms that were as thin as possible. (iii)As shown in the iext. 200 A gold films are thick enoughto be used as substrates d for forming ordered SAMs of alkanethiolates, and these gold films were also thick enough to be used as masks for the etching of underlying layers of SiOz and Si. c Measurements of 1.". The intensity of secondaryelectrons FC (1-")was measured with a SEM (JEOL-6400)by setting the mode in the line scan. Samples of Figures 1 and 2a were Ioadedinto the SEN{ chamber at the same time; readings of 1.,.were taken under the sameconditions (spot size, accelerate voltage,filament current, contrast, and brightness). Microcontact Printing (pCP). PDMS stamps were fab- 0 0,04 0.08 0.12 ricated accordingto the published procedure.l'r0A solution [KrFe(CN)r](M) of hexadecanethiolin ethanol (-2 mM) was used as the "ink" for pCP. Microcontactprinting was carried out in a clean room Figure 2. (a) Dependenct' of' the irttt'ttstties of secondar-'t- ( < 100particles/ft'r). Gold films (200A thickt usedin this paper electrons (1.,,)from both bare golcl t't'gion. if-tlled circlesi and were also preparedin a cleanroom ( < 10 000 particles/ft'r) and SAM-covered regions (open cit'clt's,ort the dr.tra.tionof'etching were used as prepared immediately without any further (for 200 A thick gold filnrs, Tht' clillt't'er.rcebetrveen these tu'tr cleaning. After application of the hexadecanethiolsolution (by intensities was plotte'das tt't:inglt's.The' quantitr' 1r is delflned - cotton Q-Tipt to the PDMS stamp, the stamp was dried in a as the time for complt'te ertching.and r1: 11)as the etching stream of Nz for -1 min and then brought into contact with windor,r'.For the composition of the standard etching snlution the surface of gold. After -5 s, the stamp was separated used (K:SzO',/O.1\'{. KOH/1.0 M. KrFe(CNriv0.01 M. and It- carefully from the gold surface. FetCNt/O.OO1 trlr, the'e'tching rvindorv is -'1 min. (b. c, dt Etching of Gold. The composition of the standard ferri/ Dependence rtf'1r and 1yon the concentrations of KzS:Or, KOH, ferrocyanideetchant was KzSzO3(0.1M), KOH (1.0 M), I(rFe- and K;FetCNtr. respectiveiy. Gold films of 200 A thickness (CN)6(0.01M), and ItFetCNro(0.001 Ml. Coution:potassium were used. When varving the concentration of each component, ferricvanide is l.ight sensitiue. The photodecontposedproducts the concentrations of other components were kept unchanged contain free cyanide.la This etching soluti,onshould be stored at the values as in the standard etching solution. The curves in brown hottles. Potassium ferricyanide is also irtcompatible were the best-fit. with acids and li.beratesHCN. All componentsof each etchant were rveighedin a chemical lab, and prepared into solution in All etchingsof gold were carried out in the clean room ( < 100 a clean room ( < 100 particles/ftjr)using HzO of 18 MA quality. particles/ft'rt using the standard compositionat room temper- 2334 Chem. Mater., Vol. 7, No. 12, 1995 Xia et al. -8 - ature unless mentioned in the text. Essentially no differences and IQFe(CNU0.001M)' tt x 7 min, /2 tt x 4 min in the quality of the samples were observed when etchings (for 200 A thick Au).le In practical applications, the were carried out with or without visible light. The etching etching window must be long enough (several minutes) beaker) was solution (300 mL of etching solution in a 400 mL to accomplish reproducible etching of samples by com- stirred at -300 rpm with a magrreticstirring bar that was -2 plete etching of the bare regions,and without unaccept- cm long. Gold frlms (-200 A thick) dissolvedcompletely in regions. 7-8 min at room temperature. The composition of the abie etching of the SAM-covered standard cyanide etching solution was KOH (1 M) and KCN Influence of the Composition of the Etching (0.1M). Solution on the Rate of Etching. Figure 2b,c shows Etching of Silicon. Anisotropic etching of silicon was the dependenceof etching time on the concentration of carried in an aqueous solution of KOH and 2-propanol at 70 each component of the ferrilferrocyanide etching soiu- (400 for - 10 "Cl mL of HzO,92 g of KOH, 132 mL of 2-PrOH) tion. All componentsbut ferrocyanide must be present min. Prior to silicon etching, the native oxide on silicon was to achieverapid and selectiveetching. removed by dipping in an unbuffered aqueous HF solution (-I7o w/w). Like the dissolution of many metals in aqueous solutions,20the etching of gold is an eiectrochemical primary steps: Results and Discussron processwith at least two (1) Metal oxidation (dissolutionl'21'22 Assays Used To Follow Etching. The etching processwas studied qualitatively and semiquantita- Auo+ 2S2O,t2:Au(S2O3)23 *e- tively by examining scanning electron micrographs (SEMs) of gold sampleswhose surfaceshad been patterned with lines of hexadecanethiolate by ,rrCP,and etched under different conditions. - * e Figure 1 shows the progression of etching in the Auo+ 2oH Au(OH)z standard solution containing SzOr2 and ferrilferrocyanide. Immediately after exposure to the etching solu- Au(OH), + 2S2O,r2: Au(S2O.,)2"+ 2OH- tion, nucleation and etching occurredpredominantly on the regions of bare gold; no etching was observed on (2) Reductionof an oxidizer: those regions derivatized with SAMs (Figure 1a). Etch- ing on the bare regions of gold seemedto be a polishing Fe(CN)^ir * e : Fe(CN),,+ process: gold was homogeneouslythinned and did not roughen during etching. During the etching of the bare regions of gold, the etchant also attacked the SAM- The rate-determining step is usually one of these two coveredregions; the attack was seen as pinholes (or electrochemicalsteps. In this instance,it appearsthat more generally, defects) on SEMs (Figure 1b). As the oxidation of gold is rate determining sincethe rate etching continued, more defects appeared in the SAM- of etching is controlledbv the concentrationsof SzOr2 (whose coveredregions (Figure 1c) and, eventuaily, all the gold (the ligand that coordinatesto Au(I)) and OH' in the SAM-coveredregions also dissolved. We assayed function is not clear but which probabiy acts as a the etching using scanningelectron microscopy (SEM): coordinating ligand or by lowering the redox potential the intensity of secondary electrons (1..) in the SEM of gold2a.za1. No difference in the rate of etching was image is approximately proportional to the amount of observedover the range of concentrationbetween 0.001 gold left on the surface when the gold films are thinner and 0.1 M for potassium ferricyanide. The addition of than the escapedepth for the secondaryelectrons from ferrocyanide to the initial etching solution had iittle Au (the images shown in Figure 1 are of secondary influenceon the rate of etching;its presence,somehow, electrons);1."decreased as gold was dissolving.li'18The greatly reduced the density of defectsthat were formed change of 1." with the duration of the etching can, in the SAM-covered regions during the etching of gold therefore, be used to monitor the etching processsemi- (seethe next sections). quantitatively. Figure 2a indicated that the magnitude Electrochemicalregeneration of Fe(III) from Fe(IItin of .I.. for bare regions of gold initially decreased as solutions is quite straightforward. For example, regen- etching proceeded,while the magnitude of 1""for SAM- erating FeCl;.;(aq)(an etchant that is used wideiy for covered regions remained essentially unchanged. 1..- making printed circuits of copper in microelectronics) (SAM) started to drop only when the samples were from FeCl2(aq)has beendemonstrated.2r'It is, therefore, overetched (that is, when the gold of the SAM-covered possible to regenerate the ferricyanide etchant used in regions was dissoiving in substantial quantity). The (that difference between these two intensities is, the (19) The times (/r - 12)for completeetching of 1000 and 1200 A contrast) increased at the beginning, reached a steady thick gold films are 29-32 and 39-'12 min, respectivelv. (rr (Figure 2a). In Figure 2a, (20) Randle.T. H. J. Chent.Educ. 1994,71.267. Mattsson,E. state to t), and then fell April, 234. (t2 - CHEMTECH 1985, tl is defined as the time for completeetching, and (21) We assumethat the gold ions producedin the etchingprocess tt) as the etching window. For the standard etching are Au(I), since the redox potential cfferricyanide is not high enough in this to oxidize Au(0) to Au(IIIt. We'aiso assume that the coordination solution used for most of the work described number for Aul i) is two, the most common coordination state for paper (KzSzOg/O.1M, KOIVI.O M, I&Fe(CN)/0.01 M, comolexesof Aut It.22The exact value of this number does not affect the discussionmade in the text. (22) Cotton, F. A.: Wilkinson, G. AduancedInorganic Chemistrv. .'50- (17) The escapedepth for the secondaryelectrons from gold is sth ed.;John Wiley & Sons,Inc.: New York, 1988,p 940. (23) * (E': + - | (8" 100 A.18 As the tongh.t".t of a surface increases, the intensity of Au- e: Au 1.69eV): Au(OH): e Au 2OH : secondaryelectrons also decreases. 0.40eVt. (18) doldstein.J.I.;Newbury, D. E';Schlin,P.;Joy, D. C.;Roming. 124) Anfelman. M. S. Ilze Encycktpedia of' ChemicaL Electntde A. D. Jr.; Lyman, C. E.; Fiori, C.; Lifshin, E' SccLnningElectron Potentials,Plenum Press: New York. 1982,p 93. (25) Microscopy and. X'ray Microanalvsi.s,2nd ed.; Plenum Press: New Hillis. M. R. Trans. 1rusf.Met. Finish.1979.57(2),73. Beyer, York, 1994;pp 69-90. S. J.: Lukes.R. NI.LrS 71-142065 710510. Monolayers of Alkaneth,roLateson GoLd Chem.tuIater.. VoL. 7. l{o. 12, 1995 2335

min) 0.001M KeFqCN)o(7 min)

b)..-*ffi CN'lO2(o5min)

i'i i.' No K4F{CN)6 (6 rnin) i,;r';:l "4i..,,. 100um 100nm Figure 3. SEMsof test patternsof gold{200 A thickrtirat ir.erefabricated by ruCPwith hexadecanethiol.fbllou'ed bl' etchingin thestandard solution of S:Or2i FetCN in ' ,'Te((lNt*: for 7 min and of CN /O2for 15 min. respectiveiy.The bright ri'gionsare goldcovered bv SAM. and the dark regionsare Si SiOrr'vher-e underivatized golci has been removed bv etching. Thc'blemishes on the dark regionsare residualgold. rhis paper (i.e., convert ferrocyanideto ferricyanide) using electrochemicalmethod-".26 Edge Resolution and Defects. Figure i3compares test patterns of gold that \\'ere fabricated by rCP of' hexadecanethiolon gold, follor,vedby selectiveetching 1pm in the standard etchants that contain S:O,l tferrtl '^ 'r f'errocvanideand cyanide. respectively. The etching Figure 4. SE\ls of test lrirtlern. ,,1- : - . r. t hrt , solution containing S.3O.r:/ferri/ferrocyanide produces u'errefabricated bv lCP of hexadt't'rrrt'tl , . : . : :, ''ri itlttg in the'solutionsthat contain 0.001 \i .'r i : :'r' r : l.litt' tOI' microfeaturesof gold with feu'er defectsand higher edge 'fhe 'il,'l' 7 and 6 min, rc-spectivel.r'. ('()rt('('r1l'r'.i.- I ('(tIll- resolution than that using c1'anide and acts morc positions rvererthe same as ttt tht' :lll, 1.,ri : :.- - ;'.1'ii()ll. rapidly. The etching behar,iorof gold in thesc tu,o etchants also appearsto be qualitativel-vdiff'erent: SEX'I regons an' Si/SrO, tr'|11';'6'ttttdt'r'tvltl rz,',i *, , . indicatesthat. in the S2O,j /ferri/ferron'anidesolution, bv ctclring. Tht' lrit'nrtshcsttn thr' ,liitl. t - etchingon bare goldis a polishingprocess (homogeneous golcl. nucleationand etching):in thc'cvanidesoli-rtion. etching on bare gold is a pitting process(heterogeneous nucle- (suchas pinholes)in the SAMs. Fltt'htnr. 1t,Lt1'-r'('iut ation and etchingt. The pitting of bare gold in the amplifvflaw-s in the SAMs. Althorr,lhth,'nriirrrl it'n-ttr cyanideetchant implies an etchingmechanism similar and size of defects\\'cr(t dt'tet'r"trttit'ri hr th,'>.\\1 rr.t'11. to the corrosionof iron in which part of the metal acts the final densitl''and siz,<'otiir'1,'t'l: |)l'lrl,r't'tl i:l >.\\l- irs an anode and part acts as a cathode.and oxvgcn is covered regions after ctchirrg iu'('tttl'lttt'nt,'tll') iitt' 1 reducedon the gold surface. The homogeneousthinning t:oordittatingligand tprobahlvI't'lltt1'rl t,r 1irr .ri:()r'tillt()ll of bare gold in the fe'rricyanideetchant indicates that constAnt of the ft:sulting coordillillIil! !',r|])r r.ll-rlttst'cl I t'erricyanideis probabll'reduced in the solution ra.thcr iri t'lich etchant: etch:rnt- ct,n1illt,ti :,irt1i't't'tttcottt'di- than on the gold surface. nating ligands usuallr.r{'::uli rr riiili'tt'tttt'lchrng mor- The addition of ferrocyanide to the initial etchrng phologres(that is. difft'r'trri tlt',r1r'1.tirt'itttcleation and solution containing S1O,r2and ferricyanide helped to grou,th of defectsr in t irt' >.\\1-r',rvt't't'clt't'girtns. reducethe density of defectsformed in the SAM-co"'ere'd Other Coordinating Ligancls. Other anions- regions(probabl-v via lowering the redoxpotential of the especiallt'thioct'ltnat('' IiS(-\ . hrilicleslKCl, KBr, KI), ferri/ferrocyanidepair). Figure 4 compareste-st patterns hypophosphatc ,Iill,I)O; . and dithiocarbamate of gold that were fabricated bv etchants with and ((CzH; )zNCS': i ( ,ll-,r , \ j'-.\\:(rt'ealso tested as coordi- rvithout addition of ferrocl,anideto the initial etching nating ligands r u:urg.ti'r'ricyar-rideas the oxidant).2; No solutions,respectively. etching was obst'r'rt'din solutions containing bromide, The defectsproduced in SAM-coveredregions during chloride. hvpoph,r.lrhll r,. and dithioczrrbamate anions. the etching of gold seem to originate from the domain Selectir.eetching rr'{isohscn'r'd for those etchants that structures of the SAMs or from the intrinsic defects

t27 r The concentration t',it'ach iigand \\'it: 0.1 ]l . the concentrations i26i Palanisamv.R.: Sozhan.G.: Narasimham.K. C. Inttian .J of'other components u'ere the sanrc iis those in the standar-d etching Tccltnol.1993. 31/,9r. 597. solution o1'ferrri. f'errocr an rde 2336 Chem.Mater.. \'ol. 7. l,io. 12. 1995 Xia et al. KzSzOslTmin

KSCN/6min 10pm ffi

1pm Figure 6. SFII'Isof a test patternof goldthat wasfabricated using thioureaetchant {pH :. l)with a patternedSAM of hexadecanethiolateas the resist.The edg'eresolution of gold - structuresthat are fabricatedusing this etchantis -0.3 rm. The bright regionsare gold coveredby SAN{,and the dark regionsare Si/SiOz where underivatized gold has been removed 5pm by etching.

Other Oxidants. Other oxidants (KClOz, KClOs, and KIO.I; all have approximately the same redox potential as ferri/ferrocyanide)were also tested as the oxidant in the etching solution, with thiosulfate or iodide as the coordinating ligand.30 No etching of gold was obsen'edrvithin -30 min. An Acidic and Selective Etchant for Gold. The I etchants for gold described so far are ail strongly alkaline solutions(pH l: 14). In someappiications (for 100nm example. rvhen the underlying substrate is GaAs), ive Figure 5. SENIsof'test patterns of gold r200 A thickr that require etchants that can function in neutral and acidic were fabricated b1' rrCP rvith hexadc'cat.iethiol.followed by solutions. We found that thiourea could be used for this etching in solutionsthat use ferricvanide'as the oxidant. and thiosulfate. thiocyanate, and iodide ion as the' coordinating purpose. Figure 6 shows SEMs of test patterns of gold ligand (all at the sameconcentration, 0.1 Mr. respectively.The that were fabricatedby aCP of hexadecanethioi,followed bright regions are gold coveredb"r, SAlt, and the dark regions by etching in an acidic etching solution (pH e: 1.tthat are Sir'SiOzwhere underivatized gold has been removed by used thiourea as the coordinatingligand and hydrogen etching. peroxide as the oxidant.3l The etching of gold in this etchant was extremel-n-rapid: bare regionsof gold films used thiocyanatel0 and iodide2s'2eas the coordinating of 500 A thick were completely etched in -30 s. The ligands. Figure 5 shows test patterns of gold that were etching window was also narrow: 5-10 s. The edge prepared by rrCP of hexadecanethiol. followed by etching resolution of this etchant is lower than that of fentJ in the etchant that used SzO:2 , SCN , and I as the ferrocyanidesolution (,'0.3 versus -0.1 trm). coordinating ligand. respectively, Thiocyanate is not as Microstructures of Gold as Masks. One of'the good as S:O,;2 as a ligand as a component of a selective applications of ,rrCPon gold is preparing goid micro- etchant: more defects are formed in the SAM-covered structures to be used as masks for the etching of regions. Iodide is as good as SzO;2-, although the time underlying substrates. Figure 7 showsSEMs of micro- for complete etching is -30% longer.

t30I The concentrationof eachoxidant was 0.01 ld: the c(Jncenrra- t28t The aqueoussolution of KI/I: has beenused to selectivelyetch tions of'other componentswere the same as those in the standard gold with thin films of photoresistsas the masks.2eThis etchant, etching -qolutioncontai r-ri ng thiosulfate. horvevc:r,had no selectivitvrvhen patterned SAMs of alkanethiolates (31t The compositionof this etchantwas 1mL of HCI (6 N), 1mL nc.reused as thc masks. of H2O2(15'i. w7wr,and 20 mL of thiourea \5%, wlv), with water as 29) Eidelloth.W.: Sandstrom.R. L. Appl. Phts.Lett. f 991.59. 1632. the solr,ent.- \:ol, Z. ^\ir. 12. 1995 2337 Monolayers of Allzanethiolates on Gold Chent. Moter., a)Au etching

iiii':lli.i

ii,i:.,i+

it:t:.i:i

i;iir.r:i .ffi'itii', b)si etching $Hf :ili.ftii llii{Jin .iiliil!+l n;{i.ildr iriiiiitii' 1n*itrr' 10pm +lll+iJ +iii{liir. ,i{*d', 'ia6s;i ilt&d), ii**tli .!s&sr l;*i{Sg ii,\qii\. dSSS: irslsF '$ttsi iilE[{{ii". .d$[tj. 'lt*$+i ,ss*:l \tffip, r$$Sr: 'it{;i+ if+i\\iii

- 10pm Figure 7. (a) SEMs of microstructuresof gold that rvere fabricatedby ,rzCPof hexadecanethiolon gold, follorvedb1' etchingof goldin the standardferricyanide solution for 7 min: tb) SEMsof microstructuresof siiiconprepared by anisotropic 10um etchingof siliconin an aqueoussoiution of KOH/I-PrOHwith I the abovestructures of gold as masks.After siliconetching, the goldmasks were removed by dippingin aquaregia for -1 min. The scalebar appliesto ali pictures.

structures of gold, and of etched silicon, that were fabricatedby a three-stageprocess: (i),rrCPof hexadecanethiolate on gold; (ii) etching of gold and chromium (iii) in SzOg2-lferclferrocyanide solution; and anisotropic * :---:-:----.-*- etching of silicon in aqueoussolutions of KOWI-PrOH.i'8 The smoothness of the silicon surface (Figure 6b) indicated that few defects were formed in the SAM- 1pm gold.16 ';-"" covered regions during the etching of itnd :tl" r.'' ''r't' Figure 8. SEMs of test patterns oi'coppel "' Selective Etching of Copper and Silver. The etch2tlt rr-ith prtttt'r-:-,t'ri:'\\l- fabricated using the ferricyantde -l'h,. ferricyanide etchant describedhere can also be used for of hexedecanethiolate {by ,rrCPr as thc rt'sisit.. i':- .ht selective etching of copper and silver with patterned regions are metals covered by SAll. ancl the dlrlk rt,{r,,ri: .:r't' SAM of hexadecanethiolate as the resist.32 Figure 8 SilSiO, rvhere underivatized metals h:rve been rt,nr,rVt.tiir\ shows, respectively, SEMs of microstructures of copper etching. and silver that were fabricated by ,lCP of hexade- removal of bare gold can canethiol, followed by etching in the ferri/ferrocyanide Using this etchant, complete interval of time with ferver solutions. The rate of etching increasesin the order Au be achievedwithin a shorter than with CN /Oz; < Ag < Cu. The times for complete etching of bare defects in the SAM-coveredregions of gold with higher regionsof these metals (500 A thick) are -17 min, -17 this etch aiso yields microstructures The micro- s, and -12 s, respectively. The selectivity and edge edge resolution than the CN /O2 system. selective resolution for Ag/CroSH are as good as those for Au/ structures of gold fabricated by rrCP follou,edby for the CroSH;while for Cu/CroSH,they are much lower. etching in the present etchant servedas masks etching of underlying layers of silicon dioxide and Conclusions silicon. The combination of S:O,r2 and ferri/ferrocyanide can also be used for seiectiveetching of copperand This work describesa more selective etchant for use silver with patterned SAMs of hexadecanethiolates(by with patterned SAMs of alkanethiolates on gold than uCP) as the resists. Thiosulfate contained in this the combination of CN lOz that was used previously; etchant can be replaced by iodide ion without loss in this SzOg2-lfern/ferrocyanide-based etching solution is the quality of the generated microstructures of gold' also less toxic and hence more convenienl 1o t.rss.l]l]'3a Acknowledgment. This work was supported in part (32) Copper and silver were etched in neutral aqueous solutions (KzSzOs/0.1M, IGFe(CNU0.01 M, I!Fe(CN)/0'001 Ml. The etching by ONR and ARPA. This work made use of MRSEC window for Ag and Cu is -15 and -5 s, respectively. shared Facilities supported by the National Science (33) Ferri- and ferrocyanideshave very high associationconstants (DMR-9400396).We would like to thank Mr' (K: 1.037,10aa,respectively);3a the concentration of free CN (formed Foundation from the dissociation of ferri/ferrocyanide) in the ferri/ferrocyanide Yuanchang Lu and Mr. Stephen Shepard for their etching solution is very low, and we believe its contribution to the assistancein using SEM, gold evaporation,and photo- etching of gold can be neglected. (34i Douglas, B.; McDaniel D. H.; Alexander, J. J. Cctnceptsand lithographic instruments. Models of Inorganic Chemistrv,2nd ed.:John Wiley & Sons,Inc.: New York, 1983,p 333. cMg50231I