Modifying of etching anisotropy of silicon substrates by surface active agents

Cent. Eur. J. Phys. • 9(2) • 2011 • 410-416DOI: 10.2478/s11534-010-0114-9

Central European Journal of Physics

Modifying of etching anisotropy of silicon substratesby surface active agents

Research Article

Krzysztof P. Rola∗, Irena Zubel

Faculty of Microsystem Electronics and Photonics, Wroclaw University of Technology,Janiszewskiego 11/17, 50-372 Wroclaw, Poland

Received 31 July 2010; accepted 25 October 2010

Abstract: The influence of alcohol additives on etch rate anisotropy of Si(hkl) planes has been studied. The etchingprocesses were carried out in 3 and 5 M KOH aqueous solutions saturated and non-saturated with alcohols.Isopropanol, 1-propanol and tert-butanol were examined. It has been showed that the etching processcannot be controlled only by the surface tension of the solution. Saturation of the etching solution withalcohols modifies etch rate anisotropy, lowering the ratio of the etch rate of (110) and vicinal planes to theetch rate of (100) plane. The morphology of Si(hkl) planes etched in 3 M KOH solution saturated withtert-butyl alcohol has been studied in detail. Smooth (331) and (221) planes have been achieved in thissolution. The (100) plane turned out to be densely covered by hillocks, opposite to the (100) plane etched inweak-alkaline solution saturated with isopropanol. To explain this phenomenon, the mechanism of hillocksformation on Si(100) surface has been proposed.

PACS (2008): 81.65.Cf, 68.03.Cd, 68.08.De, 81.16.Rf, 85.85.+j

Keywords: silicon • (hkl) planes • anisotropic etching • tert-butanol • hillocks© Versita Sp. z o.o.

1. Introduction

Anisotropic wet chemical etching of silicon is a com-mon method for manufacturing three-dimensional MEMSstructures (MEMS are electromechanical systems withmicrometric dimensions). The technology is relatively sim-ple, because it requires just a silicon substrate, a mask-ing pattern on it and an etching solution (typically KOHor TMAH). Currently, the new etching solutions are notresearched and the influence of additives applied to theetching solution is investigated. Addition of surface ac-

∗E-mail: [email protected]

tive agents (surfactants) to the etching solution lowers itssurface tension, modifying etch rate anisotropy and mor-phology of the silicon surface. Surface active agents (e.g.alcohols) are compounds containing hydrophobic tail (e.g.hydrocarbon chain) and hydrophilic group (e.g. hydroxylgroup). The addition of isopropyl alcohol (also namedisopropanol, 2-propanol, IPA) to the etching solution atthe saturation level has often been used for smootheningthe etched Si(100) surface and preventing convex cornersunderetching.

Micromirrors are one of possible applications of etchedstructures. Micromirror sidewalls of the structures can beused for changing the direction of light beam propaga-tion [1]. Depending on the crystallographic orientation of

410

Krzysztof P. Rola, Irena Zubel

the substrate and the composition of the etching solution,different angles of sidewalls inclinations can be obtained.The additives to the etching solution should be selectedin order to achieve smooth sidewall planes inclined at arequired angle towards the substrate. In the Si(100) sub-strate etched in the KOH solution with isopropanol, thesidewalls can be formed by (110) and vicinal planes. Someof them are rough after etching in this solution though [2].Therefore, to acquire smooth (hkl) planes which can formthe sidewalls, other additives to the etching solution oughtto be researched. In the last decade several alcohols havebeen studied in terms of their impact on the etching pa-rameters [3].At present using non-standard Si(hkl) substrates providesnew opportunities of shaping spatial microstructures insilicon. However, the impact which alcohols have on vari-ous (hkl) planes morphology and etch rate is different. Tobe able to control the process by surface active agentsaddition, it is necessary to understand the mechanismof additives interaction with silicon (hkl) surface duringanisotropic wet chemical etching. Usually the effect ofadditives on etching results is assigned to anisotropic ad-sorption of their molecules on (hkl) oriented silicon sur-faces [4–6]. In this model, surface active agent moleculesform a layer on silicon surface, blocking access of etchant,what results in reduction of etch rate and change of thesurface morphology. Nevertheless, it is still unknown whysurface active agents influence various (hkl) planes differ-ently.

2. Experiments

2.1. Influence of surface tension on siliconetch rate anisotropy

As it was mentioned in the introduction, addition of al-cohols to the etching solution lowers its surface tension.The influence of surface tension of aqueous KOH solutionnon-saturated with IPA on silicon etch rate anisotropy hasbeen lately investigated [5]. In this paper the researchhas been extended to other alcohols such as 1-propanol(1-propyl alcohol) and tert-butanol (tert-butyl alcohol).The sizes of molecules of the three compounds examined(presented in Fig. 1) as well as the distances between sil-icon surface atoms are similar (several tenths of nanome-ter). Therefore, it seems to be reasonable to analyze al-cohols adsorption on silicon surface at atomic scale.The research concerned 3 and 5 M aqueous KOH so-lutions non-saturated and saturated with alcohols. Thesurface tension of the solutions was measured for differ-ent concentrations of surface active agents by du Noüy

Figure 1. The properties of selected alcohols molecules (σ is sur-face tension).

Figure 2. Surface tension of etching solutions as a function of alco-hol concentration.

ring method. Fig. 2 shows an alcohol concentration de-pendence of the various solutions surface tension. The in-crease in the additive concentration in the aqueous KOH

411

Modifying of etching anisotropy of silicon substrates by surface active agents

solution causes gradual reduction of the surface tensionuntil it reaches the saturation level. While lowering thesolution surface tension, more and more alcohol moleculesgather at the solution-air interface. Similar adsorption ofsurface active agent molecules is presumed to occur at thesolid-solution interface [5].To find out how the surface tension of the etching solutionaffects etch rate anisotropy and roughness of the siliconsurface, the etching processes in various solutions withconstant surface tension (30 mN/m) were carried out. Thisrelatively low value of the surface tension is related toa high concentration of alcohol molecules in the etchingsolution, but still at the non-saturation level. Althoughalcohols probably do not form micelles as large surfactantsdo, they are believed to make an adsorbed monolayer onthe silicon surface if their concentration in the solution ishigh enough. To understand how saturation of the solutioninfluences the etching results, the etching processes werealso performed in KOH solutions saturated with alcohols.

a

b

Figure 3. Etch rates of selected Si(hkl) planes versus the angleof inclination to (110) plane: (a) in the solutions at σ =30 mN/m, (b) in the solutions saturated with alcohols.

Fig. 3a shows etch rate results of differently oriented sil-icon wafers etched in the KOH solutions non-saturatedwith alcohols. In spite of the same surface tension

(30 mN/m), the etch rate of each (hkl) plane varies in theinvestigated solutions. This indicates that the etch ratedepends more on concentrations of KOH and additives inthe solution as well as the type of the additive than on thesurface tension of the solution. Consequently, the etchingprocess cannot be controlled only by the surface tension ofthe solution. Saturation of the etching solutions modifiesetch rate anisotropy (Fig. 3b), which suggests that theincrease of number of surface active agent molecules inthe solution affects etchant behaviour until the maximumconcentration.

a

b

Figure 4. Etch rates ratio V(hkl)/V(100): (a) in the solutions atσ = 30 mN/m, (b) in the solutions saturated with alcohols:1) 3 M KOH + 1-propanol, 2) 3 M KOH + IPA, 3) 5 M KOH +1-propanol, 4) 5 M KOH + IPA, 5) 3 M KOH + tert-butanol.

The ratio of etch rates of (110), (331) and (221) planes toetch rate of (100) plane has been assessed (Fig. 4). In pureaqueous KOH solutions, the etch rate of (110) and vicinalplanes is higher than the one of (100) plane. Additionof alcohols to the etching solution changes this relation,lowering considerably the etch rate of (110), (331) and(221) planes, which results in modification of the shape

412

Krzysztof P. Rola, Irena Zubel

of etched structures, including reduction of convex cor-ners underetching and change of the sidewalls inclination.Comparison of the results obtained in non-saturated andsaturated solutions shows that the lower V(hkl)/V(100) ratiois generally in the latter ones. The lowest ratio has beenachieved in 3 M KOH solution saturated with tert-butanol.Therefore, the morphology of (hkl) planes etched in thissolution has been investigated in the next chapter.

2.2. Morphology of Si(hkl) planes etched inthe solution saturated with tert-butanol

The smoothness of (100), (311), (331), (221), (110) orientedsilicon wafers etched in the 3 M KOH solution saturatedwith tert-butanol has been studied. The morphology ofvarious Si(hkl) planes etched in weak-alkaline solutionscontaining tert-butanol has not been researched further inthe literature until now. The smoothness of (110) and (331)surfaces (which seem to be smooth in the images from anoptical microscope) has been evaluated by atomic forcemicroscope (AFM). The results presented in Fig. 5a, 5cshow the relatively high roughness of (110) plane (severaltenths of micrometer) and comparatively low roughnessof (331) plane (about 20 nm). These images are set to-gether with the images of the planes etched in 3 M KOHsaturated with IPA in Fig. 5b, 5d. The comparison ofroughness of the wafers etched in the both solutions in-dicates similar impact of isopropanol and tert-butanol onthe morphology of (110) and (331) surfaces.

The morphology of (221) plane etched in KOH solutionwith the alcohols has not been studied in detail so far inthe literature. Fig. 6 presents SEM image of V -groovesetched in Si(100) in 3 M KOH solution saturated withtert-butanol. The sidewalls of this structure are repre-sented by smooth (221) facets, which are inclined to the(100) substrate at an angle of 48.2°.

SEM images of (100) and (311) oriented wafers etched in3 M KOH solution saturated with tert-butanol are shownin Fig. 7. Both planes are covered by pyramidal structures(called hillocks) bounded by the slowest etched (111) side-wall planes. Although hillocks also exist on (311) planeetched in weak-alkaline KOH solution saturated with iso-propyl alcohol, their density on (100) surface etched inthis solution is very low [4, 6]. This suggests that theremust be different mechanisms of interaction between alco-hol molecules and Si(100) surface during anisotropic wetetching in the case of isopropanol and tert-butanol.

a

b

c

d

Figure 5. AFM images of Si(hkl) surfaces etched in 3 M KOH so-lution saturated with tert-butanol: (a) Si(110), (c) Si(331)and saturated with IPA: (b) Si(110), (d) Si(331).

413

Modifying of etching anisotropy of silicon substrates by surface active agents

Figure 6. V -grooves on Si(100) surface etched in 3 M KOH solutionsaturated with tert-butanol.

Si(100)

Si(311)

Figure 7. SEM images of Si(100), Si(311) surfaces etched in3 M KOH solution saturated with tert-butanol.

3. Model of hillocks formation onSi(100) oriented surface etched inweak-alkaline solution saturated withtert-butanolHillocks appearing on (100) oriented silicon surface dur-ing anisotropic etching in alkaline solutions are often as-signed to hydrogen bubbles which mask the surface from

the etching solution [7, 8]. However Nijdam et al. claimedthat pseudomasking by hydrogen bubbles is not proba-ble due to their instability [9]. They and other authorssuggest that the formation of hillocks could be caused bymasking of silicon surface by silicate particles, which arethe products of silicon etching [9, 10].Addition of isopropyl alcohol to the weak-alkaline etchingsolution considerably decreases in some way the numberof hillocks on Si(100) substrate. The presence of individ-ual hillocks on (100) plane may be caused by isopropanolmolecules which succeeded to adsorb on the surface and,in consequence, temporarily mask the surface from theetchant [5]. Similarly, tert-butanol molecules may per-form as temporary etching masks. The significantly big-ger number of pyramidal structures on the (100) orientedwafer etched in weak-alkaline KOH solution saturatedwith tert-Butyl alcohol could be attributed to strongeradsorption of the tert-butanol molecules than the onesof IPA.When alcohol molecules adsorb on silicon surface dur-ing the etching in alkaline solution, the intermolecularbonding between alcohol molecules and silicon surfaceoccurs. Surface bonds of silicon are mostly H-terminatedwhile etching in KOH solution, what makes the surfacehydrophobic. Therefore, a hydrophobic hydrocarbon chainof alcohol is directed towards the hydrophobic silicon sur-face and its hydrophilic hydroxyl group is directed intothe etching solution (containing H2O molecules and OH−

ions). The intermolecular attractions between hydrocar-bon atoms and silicon surface are Van der Waals forcesbetween carbon atoms of alcohol and surface atoms of sil-icon [7]. C-H group and H-terminated Si surface are bothpractically non-polar, so dispersion forces make the maincontribution to the Van der Waals attractions [11, 12].In case of isopropyl alcohol whose molecules are ad-sorbed on the silicon surface, the intermolecular bondingis formed primarily between two carbon atoms of the alco-hol which are the nearest to the surface and a few siliconsurface atoms (Fig. 8a). However, when the tert-butanolmolecules form a layer on the surface, three carbon atomsof the alcohol participate equally in the bonding (Fig. 8b).This makes the bonding stronger, what results in easieradsorption as well as more difficult desorption from thesurface. Therefore, tert-butanol molecules are more proneto adsorb and mask the surface from the etchant than onesof isopropanol. The simplified mechanism of hillocks for-mation on Si(100) surface etched in weak-alkaline solutionsaturated with tert-butanol is presented in Fig. 9.The stronger adsorption of tert-butanol molecules canbe partly confirmed by a lower surface tension of it(19.56 mN/m) than the one of IPA (21.7 mN/m). Whenthe surface tension of the solution is decreased, the hy-

414

Krzysztof P. Rola, Irena Zubel

Figure 8. Schematic presentation of alcohol molecules adsorp-tion on silicon surface in case of: (a) isopropanol(b) tert-butanol.

Figure 9. The simplified model of hillocks formation duringanisotropic etching of Si (100) in weak-alkaline solutionsaturated with tert-butanol (Si(100) surface at the begin-ning of the etching process (1) and in the end of the pro-cess (2)).

drophobic surface becomes more wettable by the solution.Nevertheless, the presented model has to be verified bycharacterization methods such as ellipsometry, contact an-gle measurements and FTIR spectroscopy. The hypothesismight be also proved by polarizable force field and molec-ular dynamics simulations. This will be the subject of thefurther research.

4. ConclusionsThe influence of alcohol additives (1-propanol, iso-propanol, tert-butanol) on silicon etch rate anisotropy hasbeen studied in this paper. It has been shown that the sur-face tension of etching solution is not a sufficient parame-ter to control the etching process. The etch rate dependsmore on factors such as concentration of the solution anda type of the additive. Saturation of the solution with thealcohol causes modification of etch rate anisotropy, reduc-ing the ratio of the etch rates of (110) and vicinal planesto the etch rate of (100) plane. The lowest ratio occursin 3 M KOH solution saturated with tert-butanol, whichresults in the most significant reduction of underetchingof convex corners [3].The morphology of Si(hkl) surfaces etched inweak-alkaline solution saturated with tert-butanolhas been investigated in detail. In contrast to iso-propanol, the impact which tert-butyl alcohol has onthe smoothness of various (hkl) planes has not beeninvestigated extensively so far. The (110) and vicinalplanes are relatively smooth after etching in 3 M KOHsolution saturated with tert-butanol. AFM and SEMimages indicate low roughness of (331) and (221) planes.The (100) surface is densely covered by hillocks, con-trary to (100) oriented wafers etched in weak-alkalinesolution saturated with IPA. Therefore, the mechanism oftert-butanol molecules interaction with Si(100) surfaceduring anisotropic wet etching has been proposed. It issuggested that tert-butanol molecules mask the siliconsurface from the etchant, which leads to hillocks formation.This model has to be proved by further investigation.

AcknowledgementsThis work has been partly supported within European Re-gional Development Fund, through grant Innovative Econ-omy (POIG.01.01.02-00-008/08) and by Wrocław Univer-sity of Technology statutory grant.

References

[1] D. Resnik et al., J. Micromech. Microeng. 15, 1174(2005)

[2] I. Zubel, M. Kramkowska, Sensor. Actuat. A-Phys.115, 549 (2004)

[3] I. Zubel, M. Kramkowska, Sensor. Actuat. A-Phys.101, 255 (2002)

[4] I. Zubel, M. Kramkowska, Sensor. Actuat. A-Phys. 93,138 (2001)

415

Modifying of etching anisotropy of silicon substrates by surface active agents

[5] I. Zubel, M. Kramkowska, Acta Phys. Pol. A 116,S-105 (2009)

[6] P. Pal et al., J. Microelectromech. S. 18, 1345 (2009)[7] M.A. Gosálvez et al., J. Micromech. Microeng. 19,

125011 (2009)[8] W. Haiss et al., J. Electroanal. Chem. 597, 1 (2006)[9] A.J. Nijdam et al., J. Appl. Phys. 89, 4113 (2001)

[10] E.A. Khizhnyak, A.V. Yukhnevich, Russ. J. Gen. Chem.77, 1315 (2007)

[11] P.C. Hiemenz, In: J.J. Lagowski (Ed.), Principles ofColloid and Surface Chemistry, Second Edition, Re-vised and Expanded (Marcel Dekker, New York, 1986)630

[12] J.N. Israelachvili, Intermolecular and Surface Forces:With Applications to Colloidal and Biological Sys-tems (Academic Press, San Diego, 1985) 73

416