Available online at wwwsciencedirect.com Sensors and Actuators A 121 (2008) 121-130 PHYSICAL ‘wow elseviercomocatna The effect of anisotropic wet etching on the surface roughness parameters and micro/nanoscale friction behavior of Si(1 00) surfaces S. Chandrasekaran, J. Check, S. Sundararajan*, P. Shrotriya Mechanical Enghwering Departmen, lwo State Cnwesy, Ames, $0011, USA Roses! 9 uly 2004; received in evised form 28 November 2004; accepted I December 2004 Availabe online 19 January 2005, Abstract Fiching processes can affect the surface roughness and hence the tribological properties of silicon surfaces. In this paper, we evaluate the surface roughness parameters and micro!nanoscale fiction behavior of Si(1 00) surfaces etched using 8M KOH and tetramethyl ammonium hydroxide (TMAH) solution with and without isopropyl alcohol (IPA) additive. Amplitude and spatial parameters were evaluated using atomic force microscopy (APM) and proflometry at scan sizes ranging from 1 to 500 ym. Results shorved that TMAH and KOH produced ‘comparable roughness up to Sy scan size and that at larger scan sizes, TMAH produced rougher surfaces than KOH. The use of IPA additive ‘cased enhancement of sub-microa roughness features as well asa sedhction i te long-range roughness of the surfaces resulting in smoother surfaces than the pure etchant. All etched surfaces exhibited pit like festures with TMAH producing slightly larger pits than KOH. Surface oughness evolution spectroscopy (SRES) showed that using TPA resulted in an increase in the maximum pit size. Single asperity fiction ‘behavior correlated well with the adhesive forces for the various surfaces—KOH and TMAH showed comparable betavior and the use of [PA resulted in loser fiction forees, However the tse of IPA resulted in surfaces with higher real area of contact, which was responsible for higher feition forces in multiple asperity contacts on the microscale. This study demonstrates tha the choice of etchants and additives affect the surface roughness and microscale frietion behavior ofthe resulting surfaces, (© 2004 Elsevier B.V. All rights reserved ‘Keywords: Sicoa; Etching: Surface roughness; Friction 1. Introduction, Surface topography and roughness are important param- eters affecting the mechanical [14] and tribological [5.6] behavior of structures and devices in microelectromechani- cal systems (MEMS). Etching processes play a critical role in the fabrication schemes of silicon based structures and devices and contribute towards the topography and rough- ness of the resulting surfaces. Several anisotropic etchants, hhave been used for etching Si surfaces such as potassium hydroxide (KOH), sodium hydroxide (NaOH), lithium hy which is usually used as a mask dur- ing the etching process. Several studies have shown that the ‘choice of etchant, etchant concentration, temperature and sil- icon erystallographie direction affects the roughness of the resulting surfaces [6,812]. The degree of anisotropy of the ‘etchants can be changed by the addition of various additives to ‘the anisotropic etching solution [13,14]. The addition of iso- propyl alcohol (IPA) to KOH and TMAK solutions was found ‘to result ina visually smooth surface compared to those from pure etchants [15]. The etching characteristics of dry etching ‘methods such as reactive ion etching (RIE) and deep reactive jon etching (DRIE) have also been studied [16,17]. Most of these studies have used optical microscopy or scanning elee- ‘won microscopy to qualitatively evaluate the roughness of the ciched surfaces, However a complete descriptive evaluation ‘of surface roughness parameters was not sought. Some stud=fey Chanrosckaran ea. / Sensors and Aerators A121 (2008) 12 ies employed a profilometer to obtain an average roughness, (R). A profilometer may not capture the entire frequency spectrum of the roughness due to large probe radius and may also cause localized damage due to high stresses [18]. Atomie force microscopy (AFM) can provide high-resolution topog- raphy image owing to very sharp probe radii, low contact stresses and sub-nanometer displacement detection systems. To the authors’ knowledge, very few studies have focused on the roughness analysis of the etched silicon surfaces using AFM. In tis paper, we present our investigations on the effect of selected etchants and additives on the surface roughness pa- rameters of the resulting etched silicon surfaces using AFM. ‘and profilometry. Using an AFM provides surface roughness information on length scales pertinent to contact regions in MEMS. Surface roughness evolution spectroscopy (SRES) ‘analysis was carried out to determine the prominent frequen- cies that arise due to the etching processes. Single asperity and ‘multiple asperity friction experiments of the various etched surfaces were also carried out to evaluate differences in tribo- logical behavior. A dry contact model incorporating surface roughness parameters was used to predict real contact area ofthese surfaces. Correlations between observed tribological, behavior and predicted real area of contact are discussed 2. Experimental details Inall the experiments, n-type silicon (1 00) waters were used. The wafers were cut into 1emx lem samples and cleaned using the RCA standard clean process [19] before etching to remove the surface contaminants. Hydrogen fluo- ride (HIF) solution was not used during this cleaning process ‘as the removal of the oxide film exposed the Si surface to ‘metallic impurities [20]. The etchants studied were TMAH. and KOH. A 25% aqueous solution of TMAH was used at, 90°C. A KOH solution with 8 M concentration was used at a ‘temperature of 80°C. The effect of adding isopropyl alcohol (PA) to the etchants on the resulting surface roughness was also studied. The solubility of IPA in KOH solution depends, ‘on KOH concentration but in TMAH solution the solubil- ity is unrestricted (15]. For our study. IPA was added to the KOH solution until saturation whereas 20% IPA by volume was added to 25% aqueous TMAH solution. A SiO» layer ‘was used as a mask to determine the etch rates of the var- ious etchants. Wafers were thermally oxidized to an oxide thickness of 300 nm and patterns were developed using pho- tolithography to expose a 0.5 em x 0.Sem square region of Si, After the etching process the SiO mask was removed by. dipping the samples in HF solution. Etch depths were then measured using a profilometer (DEKTAK 1 LA) with a probe radius of 12.5 yum. All the samples were etched to a depth, ‘of about 10 um. After etching with the anisotropic etchants the quality of the surface was generally poor due to the de- position of the reaction products on the wafer surface (21) ‘The reaction products were removed using an SPM clean 130 (H13804:H;02 = 4:1, 90 °C) followed by dipping the samples, in 37% HCl, The surfaces obtained after the cleaning pro- cess were of excellent quality with no contamination, Surface roughness of the etched Si( 0.0) samples was determined us- ing an atomic force microscope (AFM, Dimension™ 3100, Nanoscope IV, Digital Instruments/Veeco Metrology). Sur- face imaging was carried out in tapping mode under ambient conditions (25 °C, 40% RH) using a silicon tip (ip radius of | about 10nm)atscansizes of Lum x I wm, 5 wm x S pmand, 20 um x 20 pm with a resolution of 256 x 256 data points, per scan. These scan sizes correspond to typical contact and, structure sizes in MEMS and the sharp AFM tip can provide information on high frequency roughness. In order to evalu ate long-range roughness (waviness), $00 jum profiles were also obtained using the profilometer at a lateral resolution of 1am, Single asperity friction measurements were carried out using the AFM with a SisNy probe (tip radius of SOnm) in contact mode. The friction response of the tip on a sample ‘was taken to be the difference between the lateral deflee~ tion values of forward and reverse scans of a given scan line (‘e. from the friction loop of a scan line). This method is, commonly used to eliminate contributions from non-fiction sources [22]. The friction value thus noted is a measure of the friction force. The adhesive force between the SisN, tip, and the samples were measured from the force displacement, curves, All single asperity friction and adhesive force mea- surements were conducted at a humidity of 30% RH. ‘Multiple asperity friction measurements were taken with ‘aball-on-flattibometer under linear motion of the sample. A. silicon nitride ball with a 1.2 mm radius and surface rough- ness (RMS) of about 2 nm was rigidly attached to the end of a crossed I-beam structure, This was lowered using a linear stage to apply a normal load to the sample. The normal and. frictional lateral forces were measured using semiconduc- tor strain gages and were monitored simultaneously using an ADC card and a personal computer. The samples were affixed to another stage set perpendicular to the beam, providing the linear motion. Each trial was started with the sample station- ary, and an initial load of approximately 0,2 mN, The sample was then moved through a distance of 10mm at 0.6mm/s as, the load was increased to 200 mN. 3. Results and discussion Fig, 1 shows the experimental etch rates of the silicon (1.00) plane for the etchants used. The etch rates are com- parable to results obtained by other groups [7,8,11,15.21] ‘The etch rate of solutions with IPA are slightly lower than the etch rates ofthe pure solutions, This reduction in the etch, rate due to IPA has been attributed to adsorption of IPA at, sites on the crystal surface which inhibits the etching reac- tion [15]. Fig. 2 shows typical AFM topography maps ob- tained on unetched Si(1 00) and surfaces etched using vari- ous etchants at scan sizes of I um x 1 wm, 5 jum x S pum and,S Chandrascharan eal. Sento and Actuators 121 (2005) 121-130 ba zs Etch rate (ymin) oe 00 eMKOH TMAH Fig, 1. Measured etch cates of KOH and TMAH etchaats with isoprogy! aleobol (IPA) adtive and without (pu ia S100). The addition of IA, Fests in slightly lower eth ate 20m x20 um. The polished Si wafer has very low rough- ness and exhibits no dscemble features onthe surface The etchodsutacos ace visually rougher han unetched Stand ex- hibit pit ike stuctres at $ and 20 jum scan sizes. TMAH ap- perso produce ager it like sretres han KOH. Typical tlameteso these pits atthe 0 um sean sizeae 2and 3m for KOH and TMAH, respectively, whereas typical depths ane 4and 10nm, respectively. The adton of IPA esl in move high fequeney features at I pm scale: However at the 20 um scale the visual dlfeence tn the topography caused by the addition of IPA is less obvious. ‘We utilized surface roughness spectroscopy (SRES) [23] to characterize and undersiand the evolution ofthe Fourier components (spatial frequencies) of the surface roughness Caused by various etchant at the 20 um seale, According to Kim et al. [23], the function Inf/t(as, ev2, t)/he(enr, 2, 0)] fives the ecophnes evolution ofan etched surface where, Foy, 09.1) and ho. O) corespond to the Fast Fourier Transforms (FFT) of the etched surface and original Si surface respectively. As shown in Fig. 3, the functions Aer, 2,1) and hie,,a,0) were evaluated numerically using an FPT code and the APM sufce height data at 2oum. The fintion Ifh(ox,03-0)/fey-an.0)) Is ob- inind by subtracting the natural logarithm of FFT plots as Shown in Fig, 3 and numerically smocthing the data Fig. 4 Shows the experimentally measured values ofthe function Infor. es, 2/R(or< ex, 0) forthe various etched stoves, Note that the plos show thatthe addition of IPA results in a slightly Inger range of roughness evolution (the vertical Scale for the IPA pots are larger) compared 1 that of the corresponding pc tc. Fr auch etchant thor are pels aca patculer frequency (wave number) that are enhanced compared to other frequencies This suggests that features ‘with particular wavelength become prominent due to eth ing. We atempied to verily whether these peaks were re- Ite othe sizeof the prominent pt ike features seen onthe etched surfices (Fig. 2). The distance between the peas in the frequency domain is inthe order of 15-3 oder. The prominent wavelength inthe fequeney domain i calculated Ting the elation x= 32 where, xi the distance between “Table Comparison of prominent feature size generat vaiousetcantsat20 jum scan tze obtained foun AFM topography aad SRES analysis Sample Prominent enue i (a) NI SRES sxKOH 2a 2a SM KOH+IPA, 26 2s TMAH uo 2s ‘TMAH IRA 31 a ‘RES esimater ave comparable tothe average ptsies messurd fom AFM topography. The extinates also show that pt ize increases de to adition oflPa the peaks and ). the prominent wavelength, The average pit diameter also was evaluated from 2D analysis of AFM topog- raphy maps. Table 1 shows the pit size comparison between the AFM and FFT data, The numbers obtained by the two methods are quite comparable. Moreover both methods indi- ccate similar trends: (a) TMAH produces slightly larger pits as compared to KOH and (b) etching with additive results in, ‘an increase in the pit size as well. These results suggest that, SRES analysis can be used to predict prominent feature sizes that arise on an etched surface. Next, we evaluated the surface roughness parameters of the various surfaces. Surfaces with random surface height distribution and exponential autocorrelation function can be ‘completely characterized using the amplitude parameters centerline average (R,), root mean square (RMS), maximum, peak to valley distance (PTV), skewness (Sk), kurtosis (A), and the spatial parameter autocorrelation length [24]. For a profile of length [ and surface height distribution 260), the center-line average roughness (R,) isthe arithmetic mean of the absolute values ofthe vertical deviation from the ‘mean fine (mm) and is given by if, arithmetic mean of the square of the vertical deviations from, the reference line as shown below: 1 rvs’ = [ Maximum peak to valley distance (PTV) is the distance be- ‘ween the highest asperity and deepest valley of the sur- face. Skewness (Sk) represents the degree of symmetry of ‘the height distribution about the mean and is equal to the normalized third moment of the probability density funetion, ‘p@)- about the mean line’ Ry mide wo — my de Q fo . 3bt S Chandotanan ota tums t um si aM KOH TMAH TMAH 41PA Fig. 2. AFM topography images of uoetbedSi(1 00) and surfaces etched using SM KOH, SM KOH-IPA, TMAH ad TMAH +IBA at various sca sizes. The eched surface stows pt ke features at higher can sizes, The sditon of IPA appears to incresse high frequency coughoes atthe sub-micron sale fourth moment of p(2) about the mean line: ry ‘A Gaussian surface distribution has a skewness of zero and a kurtosis of three, Autocorrelation length (ACL) is a measure ‘ofrandomness of the surface and is the length over which the autocorrelation function [24] drops o I/e of its original value Surface roughness parameters depend on sampling length ‘and heace we report the variation of these parameters for a sampling length (scan size) range of 1-20 ym. Fig. 5 shows the variation of Ry as a funetion of scan size for the various etched surfaces, All roughness parameter values reported are averages of six measurements at different, locations of the sample. Error bars are +1 standard deviation, Note that data at 500 gm was obtained using the profilome- ter while the other data was obtained using an AFM. For all the samples Ry increases with scan size. This scan size dependence is commonly seen in roughness measurements and is due to large wavelength features (such as waviness) showing up in large scan sizes [25,26]. The data indicate that, the etching process results in increased roughness and that, ‘TMAH-etched surfaces have higher R, values than KOH-S Chandraseharan ea no 20 um 20 ym AFM topography map Fig, 3 Tho sequence of steps employed to obtain the sr (20pm scan siz) etched surfaces. Also the addition of IPA results in lower roughness as compared to pure etchants at $00 jm size. Fig. 6 shows the variation of RMS, PTV, skewness and kurtosis as, function of scan size for unetched Si(1 00) and the various, etched surfaces as a function of AFM scan size. RMS and 8M KOH wave number (radium) ot “15. wave number (radiyim) 15 TMAH wave number (radu) —Z “15. wave number (radium) Fig, 4 SRES maps of srfices etched using the 10 wow neta iain) & / 18 ho.040) Sonos nd Aerators 4121 2008) 121-130 oe 35 40 wavenumber (um) 40 Inth(eo.en,t/F(,0050)] roughness evolution spectroscopy (SRES) map ofan etched surface using AFM topography data PTV generally increase with scan size which is consistent with previously reported studies [26]. The surfaces etched swith the anisotropic etchants have comparable RMS values atscan sizesof | wim. Atscan sizes of 5 pmand above TMAH exhibits higher roughness than 8 M KOH. Fig. 6a and b show aM KOHAIPA. wave number (radhum) —& 38 “15. wave number (radium) 15, TMAHAIPA wave number (radum) 15 rious chants. The lighter regions represent relative stchauts appear to ence roughness ofa characterise waveauber (fequency), “15 wave number (rad/um) 35 15 higher magnitude of increase in the roughness. AlLs Chanrosckaran ea. / Sensors and Aerators A121 (2008) 12 a (om) ‘Scan size (um) Fig S. Variation of comer ine average roughness (Ry) a function ofan sizo for vanous etched surfices and unetched Si 00). Values ware ob Tsied from AFM topography’ dats (1 and 20 am) and tom proflometer ata ($00 un. ‘that at 1 um scan size, surfaces etched with IPA additive re- sults in higher RMS and PTV than surfaces etched with pure ‘etchant, This suggests that IPA acts to enhance high frequency features, which agrees with the observations of topographical 130 features at 1 um (Fig. 2). However at 5 um sizes the effect, of IPA is less clear. While KOH+ PA continue to result in rougher surfaces than KOH, TMAH + IPA begins to reduce roughness compared to TMAH. The RMS data at 20 pm and, R, data from Fig. 5 suggest that for both KOH and TMAH, IPA helps to reduce long-range roughness features resulting, in lower average roughness. This is in agreement with the qualitative conclusions reached by others via visual inspec tion of etched surfaces [15], The skewness and kurtosis data suggest that the etched surfaces are slightly non-Gaussian, We found no common trend in the variation of skewness and kur- tosis with scan size. The etched surfaces generally show more negative skewness at | and Sum scan sizes, At 20 um scan sizes, surfaces etched with IPA additive show a more positive skewness than surfaces etched with pure etchants, The sur faces show fairly comparable kurtosis values at 1 and $m. However at 20 um, TMAH +IPA etched surfaces results in a significantly lower kurtosis than TMAH-etched surfaces. In the case of KOH, IPA did not have significant effect on the kurtosis even at 20 pum scan size. Fig. 7 shows the spatial, parameter, autocorrelation length as a function of scan size. ‘The autocorrelation length gives the degree of randomness of the surface [24] and has been shown to affect contact behav~ 4 as So Ta @ «0 » BM KOHSIPA g 3 TMAH < si Be Ez dn a? : g 6 - g 20 : é 5 +—-—___——__? ee as Sean ze (im) Scan see (am) oa 40 @ @ 02 # 00. 2 i i i Eso Baz “ oa os 20 os oe oy eee Sean ize (um) sean size im) Fig 6, Varinton of surface roughness amplitude parameters a function of can size for uaetched Si and surfaces ete uring various etchant obtained fom AFM topography data: (a) RMS, (peak to valley distance, () skewness and (c) kati20 “eau Ko Stan ROH, Ae Tanta | (A ‘Autocorrelation length (um) Scan size (um) Fig, 7. Autocorrelation length a5 faction of san size for varios etched surfaces obtained rom AFM topography dts, ior of surfaces and adhesion of thin films [27]. Surfaces with very high degree of randomness have high autocorrelation length. TMAH etched surfaces shows the highest autocor- relation length while KOH etched surfaces show the lowest, autocorrelation length. The addition of IPA to TMAH results, in a less random surface whereas in the case of KOH, there isno appreciable change. ‘Next We evaluate the friction characteristics ofthe various etched surfaces. In an engineering interface, contact occurs, at numerous discrete locations, typically at asperities. The contact between a sharp AFM tip and the surface simulates single asperity contact. Fig. Sa shows the single asperity frie tion behavior ofthe various surfaces as a function of normal, load over a 20 pm scan length. We verified elastic behavior of the AFM probe-sample contact via Hertzian contact anal- ysis which showed maximum shear stress fo be well below ‘the shear strength of silicon, For all samples friction foree in- creases linearly with normal load. Surfaces etched with KOH, and TMAH exhibit the highest friction forces. Etching with IPA results in surfaces with lower friction response. Fig. 8b shows the adhesive force of the various surfaces measured from the force distance curves. The adhesive forces are a combination of the solid-solid adhesion and capillary forces, due to surface water layers. From the graphs we note that the adhesive force for surfaces etched using KOH and TMAH. fare comparable and that they are slightly higher than the un- etched Si surface. The trend of the adhesive forees for various, surfaces explains theirsingle asperity friction behavior. Etch- ing with IPA acts to reduce the adhesive forces compared to those obtained using pure etchant, thus resulting in reduced friction behavior. We also measured the friction response of the surfaces using a microtribometer with a SisN, ball. We verified an elastic contact condition by calculating the plasticity index for the contact between the ball and the various surfaces, The plasticity index (¥) for an interface predicts the degree of wa Httet Friction foree (au) © 20 40 60 60 100 120 140 160 @ "Norma oad (nN) ‘Adhesive force (oN) 8 ) KOH OIA TMA TAP Fig. . (a) Friction force (abitary mits) asa function of normal load bo- ‘eon SiN tip and various etched surfaces, measured sig an AFM. (b) ‘Adhesive force data of various surfaces obtained ram force distance curves in an AFM. Note thatthe difeences in tition babavior ofthe various surfaces cortelate fail well withthe variation of adhesive forces plasticity [24] and is given by ye v (F) even 3 where is the reduced Young’s modulus, op the equivalent asperity standard deviation and the equivalent asperity curvature of the two surfces in comact, and the harness ofthe softer materi (Si, For #0. ie deformations are Tagely elastic For al ue surfaces, we obtained en index of 014-018 indicating that our micrtribometr experiments result in elastic contacts. Fig 90 shows the coeficlont of friction obuaned for the various surfaces. Surfaces etched with IPA show a slightly higher friction response compared tosurfaces etched with the pure etchant. This contrary to the tetion behavior observed at the single aspeity contact Foran elastic eoulat, etion a the mero nnoseae ean be expressed as o ‘where ris the interfacial shear strength and tthe real area of ‘contact. In the single asperity friction experiments, 4, should ‘depend on the probe radius and not on the surface roughness, In our studies, we used the same AFM probe for all measurebs Chanrosckaran ea. / Sensors and Aerators A121 (2008) 12 Coetticient of microscale friction @ ‘SAM KOH MAPA TMAH TAKER “oa KOH <3 Taw Se aM ROHNER one ) ‘Normalized load ig 9. (a) Coefcient of acroscalefeton betwean a SiN ball and vas ‘ous etched suriaces obtained using a micrtrbometer The apparent contact ‘are 0a the order of 100 um. (b) Nomnalzed vel of contact as 8 fane- tion of wormalized load fr various etched surfaces obtained wing contact ‘mode! incorporating rourhness parameters from 20 ym AFM topography ita, Note that fora aivea los, surfaces etched with IPA rest in higher ‘alae of eoatact than those abana with pure etchant, ‘ments and verified thatthe probe radius did not change during the study via tip characterizer samples. Hence 4 can be con- sidered constant for all the surfaces for the single-asperity experiments. While studies have shown that topography can contribute to the measured friction forces in an AFM (22), these can be neglected for our surfaces since there are gener- ally no sharp changes in topography [28]. Hence variations in {fiction force indicate variation in interfacial shear strengths. of the interfaces, The single asperity friction data in Fig. 8a ‘would suggest that for a given real area of contact, the sur- faces etched with IPA additive should give lower friction force than the surfaces etched with pure etchants. However the mi- crotribometer data shows higher friction behavior for the sur- faces etched with IPA than for the surfaces etched with pure ctchants. This suggests that plays an important role on the microscale. For the loads used in the microtribometer, the ‘contact area was estimated using Hertzian theory to be on the order of hundreds of square microns. At this scale, contact «will occur at multiple asperities on the surface and hence Ay «will depend on the surface roughness parameters (29,30). 130 We utilized a contact mode! developed by Kotwal and Bhushan [31] to estimate the real area of contact for our sam= ples. This model uses the Pearson system of frequency curves based on the method of moments which provides a family of curves that can be used to generate an equation for a sur- face height distribution for which the first four moments are known. Pearson defined a criterion & given by Sk(K +3) 4OK — 38K — 6) 4K — 35) where Skand Kare the skewness and kurtosis ofthe surface as, ‘measured from topography scans. Based on the Hertzian con- tactsize, we used roughness parameters from 20 yum x 20 um, scans of the surface from Fig. 6. Depending on the value off, different equations ean be obtained for the probability den- sity function, The chosen probability density function is then usedin the classical Greenwood and Williamson's model [32] to obtain expressions for the real area of contact, number of contact spots and normal load in terms of the statistical pa- rameters of the distribution, The normal load (IP), real area of contact (le) and the number of contacts (1) is given by (31) o 4 FIAaE Ryoy Fi sth) (8) Ae = atnAaR pop Fi(h) o n= nAgFolh) (10) The standardized separation h, is given by dio where d is the separation between the mean planes of the surfaces. E” is the equivalent Young's modulus of elasticity and R,, the equivalent radius of curvature, 7, Ae and op are the surface density of asperities, nominal contact area and the equivalent standard deviation ofthe peak aspertis, respectively and Fall [i e- tetas ap 4 where p(s) is the height distribution scaled to make its stan- dard deviation unity. Fig. 9b shows the variation of normalized real area of contact as a function of normalized load for the different sur faces at 20 um scan size. The normalized load is the term [Facet te nomatizes reat are of contact the rer [ =yr4hzz | andthe normalized number of contacts is the term [4] which is obtained from Eqs. (8)-(10). It ean be seen that the normalized real area of contact increases ‘with the normalized loads. Since the real area of contact is, normalized with local surface parameters, the slight differ- cence between the various surfaces obtained from the model, ‘would be magnified in reality. The results ofthe model show. that for @ given load, surfaces etched with IPA exhibit a higher real area of contact than the surfaces etehed with pure etchant. This helps to explain why the surfaces etched with, IPA additive exhibit slightly higher friction in microscale ex-‘S Chanvaskaran ota. / Sensors and Aerators A 121 (2008) 121-130 9 periments compared to surfaces with pure etchant. These results indicate that the real area of contact effect domi- nates friction behavior of the various etched surface at the microscale, 4. Conclusions ‘This study quantifies and compares surface roughness pa- sameters of silicon surfaces etched with 8 M KOH, TMAH, with and without IPA additives and evaluates their mi- cro/nanoscale friction behavior. Results showed that although, ‘TMAH and KOH produced comparable roughness up to 5 ym, scan size, at larger scan sizes TMAH produced rougher sur faces than KOH. The use of TPA additive resulted in enhance- ‘ment of sub-micron roughness features but a reduction in the long-range roughness of the surfaces to yield smoother sur- faces than the pure etchants at scan sizes above 20 um. All the etched surfaces exhibit pit like features at 20 um sean sizes. Surface roughness evolution spectroscopy (SRES) showed that TMAH produces slightly larger pits than KOH and using IPA resulted in an increase in the maximum pit size. The sin- ‘le asperity friction behavior correlated well with the adbe- sive forces for the various surfaces. KOH and TMAH showed ‘comparable behavior and the use of IPA resulted in lower fic tion forces, However the use of IPA resulted in surfaces with, higher real area of contact in multiple asperity elastic contacts, ‘on the microscale. This real area of contact effect dominated the microscale friction bebavior resulting in higher friction, forces for surfaces etched with IPA than for surfaces etched with pure etchant. This study demonstrates that the choice of etchants and additives affect the surface roughness and microscale friction behavior of the resulting surfaces. Acknowledgments ‘The authors would like to thank Prof. Gary Tuttle at the Microelectronics Research Center at Iowa State University for his assistance in etching the samples and for engaging in several constructive discussions during the course of the ‘work, Financial support for the study was provided by a Unie versity Research Grant from the Office of the Provost and a tant from the Center for Industrial Research and Service at Towa State University References [i]. Sundararajan, B, Bhushan, T. Nama, ¥. Tomo, Mechanical prope ey measurements of nanoscale stuetires using an atomic Toes smiroscope, Ultamiceoscopy 91 (2002) 111 [2] F Ericson, 3. Schwst, Micromchanicalfnctre strength of si icon, J. Appl. Phys. 68 (1990) 580. [5] C3. Wits, A Ormeggi, M. Nachutovskib, Fracture testing of sits on microcanlever beams, J. Appl. Phys. 79 (1996) 2386, [AIT Vi L. Li, CJ. Kim, Microscale material tating of single ens talline silicon: process effets on suice morphology and tensile strongih’, Sens. Act, A: Phys. A 83 (2000) 172 S. Sundesarjan, B. Blshan, Static fiction and strfice rwghness studies of surface mieromachind eletostatic micromotors using an tome foreetretion free microscope, J. Vac. Sei. Technol A: Vac Surk Films 19 (2001) 1777, [6] R Maboudian, RT. Howe, Critial review: adhesion in surface mic sromechanical structures, J. Vic. Sei. Technol. Br Mroeleeton Nanometer Stuct 15 (1997) 1 (7) ML. Madou, Fundamentals of Micofibricstion,2ad od, CRC Press, New York, 2002 [8] K. Sito, M. Shika, T. Yamashiro, Ko Aswumi, Y, lve, M. ‘Yunameio, Ansouopic etching rats of siglecrystl silicon for ‘TMAH water solution asa function of enystallogaphic orientation Sens. Actua. A: Phys. A 78 (1999) 131 [9] Mo Shikida, K. Safe, K. Tokowo, D. Uchikawa, Difernees ia auisotopie etching propetes of KOH and TMAH solutions, Ses. ‘Actuat. A: Pays. A 80 (2000) 173 [No] N. Miki, SIM. Spearing, Effect of annoscale surface roughness om bonding enersy of divet-boaded silicon wafers, J. App. Phys. 4 (2003) 600. [11] K. Sato, M. Sbkida, Yamashiro, M. Tameka, Sto, Rvahen- ing of single-crystal silicon surfce eiched by KOH water soliton, Sens. Actua A! Phys. 78 (1999) 122. [12] 1 Zobel, Silicon anisotropic etching in alkaline soltons. I. On the influence of anisotropy on the smoothness of etched surfaces, Sens. ‘Actua A! Pays. A 70 (1998) 260 [15] HG. Linde, LW. Austin, Catalytic coael of anisotropic silicon etching, Sens. Actst. A: Phys 49 (1995) 181 [14] C: Moldovan, Ros, D. Dascals, G, Nechitor, Anisotropic etching of silicon in' complexant redox alkaline system, Sans. Acs. B Chom. 58 (1999) 438, [15] 1 Zabol, M. Krakow, The oft of isopropyl slcobol om tching rate and roughness of (100) Si surfice etched in KOH and TMAH soons, Sens. Actua A: Phys. A 98 (2001) 138, [6] CC. Lee, W. Hat, Method on surfice roughness modiicstion to alleviate sition of microstrcurs, 1. Vac. Sci Techaol. Bi Mic troelecron. Nanonteter Strut, Process. Meus. Phenom. 21 (2008) 1308, [17] S. Chandrasekaran, S. Sundarrajan, Elfet of microfabrication pro- cesses on surface roughness partes of silicon surfaces, Su. Coat. Technol, 1881189 (2004) $81 [1s] C-. Poon, B. Bhushan, Comparison of susface roughness measure ‘ment by stylus prole, AFM and non-contact optical profiler, Wear 190 (1998) 76 [U9] GK. Celle, D.L. Bact LM, Rosalia, icing of silicon by the RCA Standard Cleaa 1, Elecuochem. Solid-State Let. 3 (2000) 47 [20] R. Dias, N. Moldovea, H. Cmion, Roughening and smohiag dy namics doing KOH silicon etching, Sens. Actua. A: Phys. A 74 (1999) 18, [U1 1, Zabel, . Baryeka, Silicon anisouopic etching in alkaline stu ‘ons. I. The geometric description of figures developed under etch- fing Si{.00) in cis soltions, Seas. Actua. A: Pays. A 70 (1995) 250, BB. Bhushan, Handbook of MicroNauo Tribology, 2a ed, CRC Press, 1999, [25] KS. Kim, 1A. Honado, H. Tan, Evolution of a surface-oughness spoctum causa by ses ia aanometersalechemieal etching, Phys ev. Lett 83 (1999) 3872 [24] B. Blnshan, Principles and Applications of Tribology, Wiley, NY, 19. [25] R Sayles, TR, Thomas, Surice topography as non-stationary ean dom process, Nature 271 (1978) 431 [6] VN. Koinkat, B. Bhushan, Efect of scan size and surice rough- ness on microscale friction measurements, J. Appl. Pays. 81 (1997) 2472, bebo S Chandrasekaran ota / Sensors ane Actuators A 121 (2008) 124-130 [27] G. Palasanizas, LEM.D. Hosson, Inftuence of surface roughness on the adhesion of elastic films, Phys. Rev. E (Stat. Nonlinear Soft ‘Matte Phys.) 67 (2003) 021604 [28] S. Sundararjan, B. Bhushan, Topography-inoced contiations to Teton frees measured using an atomic fore'ition force mixo- scope, J. Appl. Phys. $8 (2000) 4825, [29] TR Thomas, Rough Surfaces, 2a ed, Imperial Celle ress, Lon- doa, 1999, [30 EP. Bowden, D. Tabor, The Friction an Lubrication of Solids, Pat 1, Clarendon Pres, Oxford, 1980 [51] CA. Kotwal, B. Bhushan, Contact analysis of won-Gausian surfaces {br minimum static and Liveticfston and wear, Tebol. Tras. 39 (1996) $90, [52] LA. Greenvood, 18P. Willamson, Contact of aominally Bat su Tees, Proc. Roy. Soc. Lond. A 295 (1966) 300, Biographies Sharath Chandrasekaran ws hor ia 1980 in Baagalos, India. He received a BE degres in Mechanical Engineering from Bangalore Unive Sty, India in 2001 and an MS depee in Mechanical Engineering at Towa State University in 2004, At Town State, he wilized SPM-basd metiods| to study nanoscale tology and mechanics of materials for MEMS. He is cumenly working as a research enginsor at Novsecan Technologies, ‘Ames, 1A, ‘Jason Check recived 9 BS in Physics in 1998, and is curently fnsh- ing a MS in Mechinical Engietring tfom Towa State University, His research interests are in the miroscale tribological peopetes of mate- Fils, and more specifically of biomedical materials used in tt joint, replacement. Sriram Sundararajan isan Assistant Professor of Mechanical Engines sng at Towa State Univesity. He was bora in 1974 in Choana, India and rveeived « BE degree in Mechanical Engioesing ffom Bila Tasiute of ‘Technology and Seieace, Pian (Inia) in 1995, He obaind his MS and PD depres in Mechanical Engineering fom The Ohio State Univer- sy, Columbus, OF in 1997 and 2001, respectively Since joining Towa State Unversity in January 2002, his research program has focused on imirolnescal trbology and mechanics as well as developing ssanning probe microscopy based methods to probe surface and material proper tis atthe nanoscale, He is # mtaber of ASME, STLE, ASM, MRS abd Acs, Panay Shrotriya isan Assistant Professor in Mcanical Engineering Department at Tova State University. Before joining Jowa State Univer sity infill 2008, be was a postdoctoral research associate in Division of Eagiseering at Brown Univesity from July 2002 to August 2003 and In Mechanical and Aerospace Engineering at Princeton Univesity from December 2001 to June 2002. He seceried his PAD aad MS depress in Thaorsticl and Applisd Mechanics ftom the University of Minot ‘Urbana-Champaign, His cesearch interests ain mechanics of nano- structured mens, surface roughness evolution duc to stess-assised chemical reactions and techniques for nanoscale mechanical characteri-