Reaction Mechanism, Bonding, and Thermal Stability of 1-Alkanethiols Self-Assembled on Halogenated Ge Surfaces

DOI: 10.1021/la904864c 8419Langmuir 2010, 26(11), 8419–8429 Published on Web 04/30/2010

pubs.acs.org/Langmuir

© 2010 American Chemical Society

Reaction Mechanism, Bonding, and Thermal Stability of 1-Alkanethiols

Self-Assembled on Halogenated Ge Surfaces

Pendar Ardalan,† Yun Sun,‡ Piero Pianetta,‡ Charles B. Musgrave,†,§ and Stacey F. Bent*,†

†Department of Chemical Engineering, Stanford University, Stanford, California 94305, and‡Stanford Synchrotron Radiation Lightsource, Menlo Park, California 94025. §Current address:

Department of Chemical & Biological Engineering, University of Colorado at Boulder, Boulder, Colorado 80309

Received December 24, 2009. Revised Manuscript Received February 10, 2010

We have employed synchrotron radiation photoemission spectroscopy to study the reaction mechanism, surfacebonding, and thermal stability of 1-octadecanethiolate (ODT) self-assembled monolayers (SAMs) at Cl- and Br-terminated Ge(100) surfaces. Density functional theory (DFT) calculations were also carried out for the same reactions.From DFT calculations, we have found that adsorption of 1-octadecanethiol on the halide-terminated surface viahydrohalogenic acid elimination is kinetically favorable on both Cl- and Br-terminated Ge surfaces at roomtemperature, but the reactions are more thermodynamically favorable at Cl-terminated Ge surfaces. After ODTSAM formation at room temperature, photoemission spectroscopy experiments show that Ge(100) and (111) surfacescontain monothiolates and possibly dithiolates together with unbound thiol and atomic sulfur. Small coverages ofresidual halide are also observed, consistent with predictions by DFT. Annealing studies in ultrahigh vacuum showthat the Ge thiolates are thermally stable up to 150 �C. The majority of the surface thiolates are converted to sulfideand carbide upon annealing to 350 �C. By 430 �C, no sulfur remains on the surface, whereas Ge carbide is stable toabove 470 �C.

1. Introduction

Germanium is being investigated for integration into silicon-based electronic devices mainly due to its high bulk electron andholemobilities and the lower thermal budget required for process-ing, such as lower dopant activation temperature (∼400-500 �C),that will allow formation of shallow junctions.1-3 One of themajordrawbacks to the use of Ge is the difficulty in growing an insula-ting oxide comparable to SiO2 in Si technology. Studies of Geoxide suggest that GeO2 is not suitable as a gate dielectric;4 thatGeO, while more stable thanGeO2,

5 is unstable; and thatGeO2 iswater-soluble.6 Also, upon annealing, GeO2 transforms to GeO,which then desorbs from the surface at ∼420 �C,7 and GeO gasacts as reducing agent which can negatively affect the electricalproperties of a device.3 Moreover, removal of the germaniumoxide alone is not sufficient for good electrical performance indevices such asmetal-oxide-semiconductor field-effect transistors(MOSFETs).8Forexample,Ge surfacecleaning inultrahighvacuum(UHV) conditions results in an oxide-free surface; however, sub-sequent deposition of high-κ dielectric materials was shown toresult in leaky devices, and consequently, use of an appropriate

passivation layer is necessary.9 Moreover, surface passivation ofGe has a direct effect on the quality and characteristics of theinterfacial layer between the high-κ dielectric material and the Gesubstrate.2,3,8-11

Several different passivation methods have been explored forgermanium, and a number of fundamental studies investigatingthe passivation and organic functionalization of Ge substratesunder gas phase or UHV conditions have been reported.1,5,12-21

Furthermore, various solution-based methods have been studiedto achieve chemical passivation of the germanium surface prior to

*Towhomcorrespondence should be addressed. E-mail: [email protected].(1) Loscutoff, P. W.; Bent, S. F. Annu. Rev. Phys. Chem. 2006, 57, 467–495.(2) Misra, D.; Garg, R.; Srinivasan, P.; Rahim, N.; Chowdhury, N. A. Mater.

Sci. Semicond. Process. 2006, 9, 741–748.(3) Kamata, Y. Mater. Today 2008, 11, 30–38.(4) Rivillon, S.; Chabal, Y. J.; Amy, F.; Kahn, A. Appl. Phys. Lett. 2005, 87,

253101.(5) Onsia, B.; Conard, T.; De Gendt, S.; Heyns, M.; Hoflijk, I.; Mertens, P.;

Meuris, M.; Raskin, G.; Sioncke, S.; Teerlinck, I.; Theuwis, A.; Van Steenbergen,J.; Vinckier, C. Solid State Phenom. 2005, 103-104, 27–30.(6) Prabhakaran, K.; Ogino, T. Surf. Sci. 1995, 325, 263.(7) Prabhakaran, K.; Maeda, F.; Watanabe, Y.; Ogino, T. Appl. Phys. Lett.

2000, 76, 2244–2246.(8) Caymax, M.; Elshocht, S. V.; Houssa, M.; Delabie, A.; Conard, T.; Meuris,

M.; Heyns,M.M.; Dimoulas, A.; Spiga, S.; Fanciulli,M.; Seo, J.W.; Goncharova,L. V. Mater. Sci. Eng., B 2006, 135, 256–260.(9) Soe, J.W.; Dieker, C.; Locquet, J.-P.; Mavrou, G.; Dimoulas, A.Appl. Phys.

Lett. 2005, 87, 221906.

(10) Chui, C. O.; Kim, H.; McIntyre, P. C.; Sarasawat, C. IEEE Electron DeviceLett. 2004, 25, 274–276.

(11) Shang, H.; Frank, M. M.; Gusev, E. P.; Chu, J. O.; Bedell, S. W.; Guarini,K. W.; Leong, M. IBM J. Res. Dev. 2006, 4/5, 377–386.

(12) Schnell, R. D.; Himpsel, F. J.; Bogen, A.; Rieger, D.; Steinmann, W. Phys.Rev. B 1985, 32, 8052–8056.

(13) Roche, J.; Ryan, P.; Hughes, G. J. Appl. Surf. Sci. 2001, 174, 271–274.(14) Weser, T.; Bogen, A.; Konrad, B.; Schnell, R. D.; Schug, C. A.; Steinmann,

W. Phys. Rev. B 1986, 35, 8184–8188.(15) Gothelid, M.; LeLay, G.; Wigren, C.; Bjorkqvist, M.; Karlsson, U. O. Surf.

Sci. 1997, 371, 264.(16) Ardalan, P.; Davani, N.;Musgrave, C. B. J. Phys. Chem. C 2007, 111, 3692–

3699.(17) Filler, M. A.; VanDeventer, J. A.; Keung, A. J.; Bent, S. F. J. Am. Chem.

Soc. 2006, 128, 770–779.(18) Cullen, G.W.; Amick, J. A.; Gerlich, D. J. Electrochem. Soc. 1962, 109, 124.(19) Fouchier, M.; McEllistrem, M. T.; Boland, J. J. Surf. Sci. 1997, 385,

L905–L910.(20) Bachelet, G. B.; Schl€uter, M. Phys. Rev. B 1983, 28, 2302.(21) Cao, S.; Tang, J. C.; Shen, S. L. J. Phys.: Condens. Matter 2003, 15, 5261–

5268.(22) Sun, S.; Liu, Z.; Lee, D.-I.; Peterson, S.; Pianetta, P.Appl. Phys. Lett. 2006,

88, 021903.(23) Sun, S.; Sun, Y.; Liu, Z.; Lee, D.-I.; Pianetta, P. Appl. Phys. Lett. 2006, 89,

231925.(24) Deegan, T.; Hughes, G. Appl. Surf. Sci. 1998, 123/124, 66–70.(25) Lu, Z. H. Appl. Phys. Lett. 1996, 68, 520–522.(26) Kim, J.; McVittie, J.; Saraswat, K.; Nishi, Y. ECS Trans. 2006, 3, 1191–

1196.(27) Kim, J.; McVittie, J.; Saraswat, K.; Nishi, Y. In 8th International Sympo-

sium on Ultra Clean Processing of Silicon Surfaces, Antwerp, Belgium; Mertens, P.,Meuris, M., Heyns, M., Eds.; Trans Tech Publications: Switzerland, 2006.

(28) Kim, J.; Saraswat, K.; Nishi, Y. ECS Trans. 2005, 1, 214–219.

8420 DOI: 10.1021/la904864c Langmuir 2010, 26(11), 8419–8429

Article Ardalan et al.

high-κ deposition including hydrohalogenic acids,4,5,8,22-36 hydro-gen peroxide,4,5,7,28,30,33,36 deionized (DI) water,4,6,28,30 and ammo-nium hydroxide.28,30,33,34 Deposition of oxynitride,2,8-11 oxysul-fide,37 and silicon38 layers has also been utilized forGe passivation.Our group and others have previously demonstrated that hydro-halogenic acid treatment is one of themost promising approachestoGe surface passivation.Within this class of acids, HBr andHClare better suited for wet functionalization than HF due tocomplete removal of Ge suboxides and higher stability of thepassivated surfaces against reoxidation.24,26,27,29,30,35,36

Unfortunately, since evenhalide termination produces surfaceswith limited stability, other passivation methods must be explo-red. One possibility is to attach an organic group to the surface.For example, deposition of organics by hydrogermylation39,40

and thiolation36,41,42 have been utilized for Ge passivation. Passi-vation using self-assembledmonolayers (SAMs) of thiols (thiolation)is promising due to the advantage of incorporating a Ge-S bondat the interface. The understanding of the Ge-S bonding comesfrom a number of fundamental studies done in UHV or solution;such studies have shown promising characteristics such as highambient stability (on order of a few days) for S-terminated Gesurfaces.13,14,31,37,43,44 Apart from passivation, an alkanethiolateSAM on Ge has potential application in area selective atomiclayer deposition (ASALD),45molecular electronics devices,46 bio-logical sensors,47 and microelectromechanical systems (MEMS).42

In contrast to the extensive literature on the 1-alkanethiolate/metal system,46,47 the use of 1-alkanethiolates on semiconductorsurfaces such as GaAs48-52 and Ge36,41,42 has been less studied.Maboudian and co-workers demonstrated for the first time thatwell-packed 1-octadecanethiolate SAMs covalently bound to thesurface through Ge-S bonds can be formed at the H-terminated

Ge(111) surface at room temperature andare air-stable up to12h.42

The adsorption kinetics of 1-alkanethiolate SAMs at H-terminated Ge(111) were also studied and indicate that theadsorption mechanism involves a two-step process that dependson the concentration and chain length of the 1-alkanethiols.41

Our group has recently demonstrated formation of 1-octa-decanethiolate (ODT) SAMs on Cl- and Br-terminated Ge(100)andGe(111) surfaces from solution.36 The advantage of this routeof halogenation followed by thiolation is that the initial surfacetreatment (HCl or HBr) not only removes the oxide and con-tamination but also results in better stability, making the passi-vated surface suitable for wet functionalization. By employing acombination of analytical methods such as X-ray photoelectronspectroscopy (XPS), Auger electron spectroscopy (AES), andinfrared (IR) spectroscopy, we showed in our previous studythat ODT SAM formation depends upon concentration of the1-octadecanethiols, choice of solvent, crystallographic orientationof the substrate, and type of surface passivation. A comparison ofthe Ge(100) andGe(111) surfaces revealed that ODT SAMs formwith higher surface coverage, better packing, and better ambientstability at the halide-terminated (100) surfaces. Also, a compa-rison between the thiolated Ge:Br and Ge:Cl surfaces revealedthinner ODT films but with higher ambient stability for theGe:Brsamples, although this difference is more pronounced at the (111)surfaces.36

However, the previous study did not determine the bondingstructures in detail nor did it elucidate the attachmentmechanism.The major limitation of that study was the use of conventionalXPS for surface characterization, which did not allow for analysisat sufficiently high resolution to distinguish between differentstructures. For example, the shape of the Ge(3d) fine scan peakscorresponding to thiolated Ge suggested a complex convolutionof peaks, but due to limitations in sensitivity and resolution thepeaks could not be completely resolved. Also, the low signal-to-noise ratio as well as interference of surface plasmons from theGe(3p) core level orbital did not allow us to probe the bondingusing the S(2P) peaks in the XP spectra.36 Furthermore, we wereunable to quantify the residual halogen concentration which maybe present at the surface even after thiolation.36

Hence, in the current work, the details of surface bonding atroom temperature for the ODT SAMs formed on various halo-genated surfaces were studied by employing synchrotron radia-tion. Compared with conventional XPS, synchrotron radiationphotoemission spectroscopy (SR-PES) has much higher resolu-tion and surface sensitivity due to its tunable photon energy andhigh X-ray intensity. Using SR-PES, we will show that thiolated(100) and (111) surfaces of Ge are mainly characterized by thepresence of monothiolates and possibly dithiolates as well as byunbound thiol and atomic sulfur at room temperature. To com-plement the experimental studies, we carried out DFT calcula-tions to probe the reaction mechanism and surface bonding of1-alkanethiols at halogenatedGe surfaces. These calculations willshow that hydrohalogenic acid elimination reactions are kineti-cally favorable on these surfaces at room temperature, and thatthe reactions are thermodynamically more favorable at Cl-termi-nated Ge surfaces than at Br-terminated Ge surfaces.

In addition to room temperature experiments, thermal anneal-ing studies were also performed. Such studies are importantbecause if 1-alkanethiolate SAMs formed on Ge surfaces are tobe used in electronic devices, theywill be subjected to various clean-ing and annealing steps. We will show that Ge-thiolates arethermally stable up to 150 �C and the majority of the surfacethiolates are converted to sulfide and carbide via S-C bondscission upon annealing to 350 �C.

(29) Lu, Y. F.; Mai, Z. H.; Song,W. D.; Chim,W. K.Appl. Phys. A:Mater. Sci.Process. 2000, 70, 403–406.(30) Kim, J.; Mcvittie, J.; Sarasawat, K.; Nishi, Y. Study of Germanium Surface

for Wet Cleaning Applications, SRC/SEMATECH Engineering Research Centerfor Environmentally Benign Semiconductor Manufacturing; teleconference,16 May 2006.(31) Bodlaki, D.; Yamamoto, H.; Waldeck, D. H.; Borguet, E. Surf. Sci. 2003,

543, 63–74.(32) Cao, S.; Tang, J. C.; Wang, L.; Zhu, P.; Shen, S. L. Surf. Sci. 2002, 505, 289.(33) Okumura, H.; Akane, T.; Matsumoto, S. Appl. Surf. Sci. 1998, 125, 125–128.(34) Hovis, J. S.; Hamers, R. J.; Greenlief, C. M. Surf. Sci. 1999, 440,

L815–L819.(35) Ardalan, P.; Pickett, E. R.; Harris, J. S., Jr.; Marshall, A. F.; Bent, S. F.

Appl. Phys. Lett. 2008, 92, 252902.(36) Ardalan, P.; Musgrave, C. B.; Bent, S. F. Langmuir 2009, 25, 2013–2025.(37) Frank,M.M.; Koester, S. J.; Copel,M.; Ott, J. O.; Paruchuri, V. K.; Shang,

H.; Loesing, R. Appl. Phys. Lett. 2006, 89, 112905.(38) Wu, N.; Zhang, Q.; Zhu, C.; Chan, D. S. H.; Li, M. F.; Balasubramanian,

N.; Chin, A.; Kwong, D.-L. Appl. Phys. Lett. 2004, 85, 4127–4129.(39) He, J.; Lu, Z. H.;Mitchell, S. A.;Wayner,D.D.M. J. Am. Chem. Soc. 1998,

120, 2660–2661.(40) Choi, K.; Buriak, J. M. Langmuir 2000, 16, 7737–7741.(41) Kosuri, M. R.; Cone, R.; Li, Q.; Han, S. M.; Bunker, B. C.; Mayer, T. M.

Langmuir 2004, 20, 835–840.(42) Han, S.M.; Ashurst, W. R.; Carraro, C.;Maboudian, R. J. Am. Chem. Soc.

2001, 123, 2422–2425.(43) Maeda, T.; Takagi, S.; Ohnishi, T.; Lippmaa, M. Mater. Sci. Semicond.

Process. 2006, 9, 706–710.(44) G€othelid, M.; LeLay, G.; Wigren, C.; Bj€orkqvist, M.; Rad, M.; Karlsson,

U. O. Appl. Surf. Sci. 1997, 115, 87–95.(45) Chen, R. SurfaceModification for Area Selective Atomic LayerDeposition

on Silicon and Germanium. Ph.D. Thesis, Stanford University, 2006.(46) Love, J. C.; Estroff, L. A.; Kriebel, J. K.; Nuzzo, R. G.; Whitesides, G. M.

Chem. Rev. 2005, 105, 1103–1170.(47) Ulman, A. Chem. Rev. 1996, 96, 1533–1554.(48) Voznyy, O.; Dubowski, J. J. J. Phys. Chem. B 2006, 110, 23619–23622.(49) Jun, Y.; Zhu, X.-Y.; Hsu, J. W. P. Langmuir 2006, 22, 3627–3632.(50) McGuiness, C. L.; Blasini, D.; Masejewski, J. P.; Uppili, S.; Cabarcos,

O. M.; Smilgies, D.; Allara, D. L. ACS Nano 2007, 1, 30–49.(51) McGuiness, C. L.; Shaporenko, A.; Mars, C. K.; Uppili, S.; Zhanikov, M.;

Allara, D. L. J. Am. Chem. Soc. 2006, 128, 5231–5243.(52) McGuiness, C. L.; Shaporenko, A.; Zhanikov, M.; Walker, A. V.; Allara,

D. L. J. Phys. Chem. C 2007, 111, 4226–4234.

DOI: 10.1021/la904864c 8421Langmuir 2010, 26(11), 8419–8429

Ardalan et al. Article

2. Experimental and Computational Details

Ge samples were cleaved from 2 in. Czochralski (CZ) grownn-type Ge(100) and Ge(111) wafers (Umicore, Belgium) withresistivity ranging from 1.0 to 3.0 Ω-cm. All reagents were usedas purchased, including hydrobromic acid (Aldrich, reagent grade,48 wt %), hydrochloric acid (EMD Chemicals Inc., ACS grade,37wt%),methanol (EMD,ACSgrade), chloroform (EMD,ACSgrade), hydrogen peroxide (EMD, ACS grade, 30 wt % in H2O),2-propanol (Fisher, ACS grade), and acetone (Fisher, ACS grade).Prior to use, 1-octadecanethiol (96%, Acros Organics) was driedin a desiccator and used without further purification.

Particles and adventitious organics at the as-received Ge sur-faces were first removed by 10 min sonication in neat acetonefollowed by chloroform. The samples were subsequently blowndry with N2. For halogenation, we have employed the cyclic oxi-dation and etching methodology, since we have previously foundthat ODT SAMs formed after such treatment showed higherambient stability than those formed on the surface after directdipping in hydrohalogenic acids.53 Consequently, the surfaceswere first oxidized byH2O2 (30 wt% inH2O) for 5min, and aftera DI water rinse1,4,6,28 Ge samples were etched in aqueous HClsolution (10 wt %) or HBr solution (10 wt %) for 10 min. Theseoxidation-rinsing-etching cycles were repeated three times33 toremove traces of the oxide and achieve halide termination of theGe surfaces.24,26,27,29,30After the final cycle, the halide-terminatedsamples were directly blown dry with N2.

23,27 To form the SAMs,halide-passivated samples were dipped immediately after passiva-tion into 0.1 M solutions of 1-octadecanethiol (ODT) in 2-propanol (isopropanol, IPA) for 48or72h.36All of theODTsolu-tions were freshly prepared before the experiments. To eliminatesolvent evaporationand exposureof the reactionmedia to ambient,the SAM containers were sealed by Parafilm and stored in an air-purged glovebox. After the ODT SAM formation, the sampleswere removed from the glovebox and sonicated in neat IPA for1 min to remove possible physisorbed molecules from the Ge sur-face. All samples were blown dry withN2 before characterization.

We have previously carried out water contact angle (WCA)measurements, ellipsometry, and Fourier transform infrared(FTIR) spectroscopy measurements on SAMs formed using themethod described above.36 Those results indicate that well orde-redSAMsaremadeby this procedure. For example,FTIRspectros-copy shows that the C-H stretching peaks observed for bothGe(100):Br and Ge(100):Cl substrates at 2918 and 2848 cm-1

coincide well with the CH2 antisymmetric and symmetric stretch-ing vibrationalmodes, respectively.Moreover, thepeakat 2959 cm-1

is characteristic of the CH3 antisymmetric stretching vibra-tional modes. These three principal peaks are in agreement withpresence of ordered, crystalline-like, and well-packedmonolayerson solid surfaces.54,55 For the case of Ge(111), these peak posi-tions were slightly blue-shifted, which along with lower WCAvalues suggests that ODT SAMs with less surface coverage andpoorer packing form on halogenated Ge(111) surfaces than onGe(100).

The photoemission experiments were conducted at beamline10-1 (a wiggler beamline) at the Stanford Synchrotron RadiationLightsource (SSRL).The chamberbasepressurewas 1�10-10Torr.A PHI model 10-360 hemispherical capacitor electron energyanalyzer and an Omni Focus III small area lens were mountedon the analysis chamber. To optimize the surface sensitivity andincident beam intensity, a 300 eV photon energy was selected forthe S(2p), Cl(2p), and Br(3d) core levels, while the Ge(3d), C(1s),and O(1s) were monitored at 200, 350, and 620 eV, respectively.The fine scans were taken at a pass energy (PE) of 11.750 eV.The energy resolution at 200 eV is approximately 0.3 eV. For the

annealing experiments, the samples were radiatively heated ata range of temperatures from room temperature to 470 �C for20min inside theUHVchamber.Next, the sampleswere cooled toroom temperature (RT) inside the chamber for 10min before newspectrum acquisition. The temperature was probed by using anAl-Cr thermocouple, and the temperature uncertainty is app-roximately 20 �C.

The photoemission data was processed using a Shirley back-ground correction56 followed by fitting toVoigt profiles such thata minimum number of meaningful peaks were employed for thepeak fitting. Key fitting parameters for Ge(3d) scans are 190meVLorentzian width, 280 meV Gaussian width with higher valuesallowed for þ3 and þ4 oxidation states, 0.585 eV spin-orbitsplitting,13,14,22 and 0.667 branching ratio.13,14,22 The S(2p) peakswere fitted to Voigt peaks with the same full width at half-maximum (FWHM) using 1.2 eV spin-orbit splitting,57,58 anda branching ratio of 0.5.57,58 All peaks were adjusted using thebulkGe(3d) peakat the same energy to correct thekinetic energiesfor the charge shift.59 The Cl or Br coverage at the thiolated Gesurfaces (θ) was calculated from the photoemission data based onthe exponential absorptionmodel suggestedbyRankeand Jacobi60

(see the Supporting Information).The Ge9H12X2 (X=Cl or Br) one-dimer clusters were used to

model the reactivity of the halide-terminated Ge(100) surfacestoward the 1-alkanethiol molecules (see Figures 2 and 3 for thestructures of the clusters). These clusters consist of two Ge atomsrepresenting the surface dimer with each dimer atom terminatedby one halogen atom, and seven Ge atoms modeling three layersof subsurface bulklike atoms. The dangling bonds of the subsur-face atoms are terminated by 12 hydrogen atoms to mimic the sp3

hybridization of bulk Ge. Clusters have been used previously asmodels of surface reactive sites to predict reaction products onboth the Ge and Si surfaces and generally produce results con-sistentwith experimental observations except in caseswhere signi-ficant interactions extend beyond the edge of the cluster.17,61-64

To minimize aphysical distortions of the cluster, the positions ofthe terminating H atoms were fixed following optimization of theunconstrained Ge9H12 cluster.

Calculations were performed using density functional theory(DFT) with the Becke3 Lee-Yang-Parr (B3LYP) hybrid andgradient corrected exchange functional.65-68 The electronic struc-ture was expanded over atomic Gaussian basis functions using amixed basis set schemewith the polarized double-ζ 6-31G(d) basisused for chemically active atoms (Ge dimer and halogen atoms),the 6-31G basis set used for the chemically inactive terminatingH atoms of the Ge9H12X2 clusters, and the LANL2 effective corepotential and valence double-ζ basis set (LANL2DZ) employedfor describing the chemically inactive subsurface Ge atoms. Wehave employed a truncated model (1-ethanethiol) to describe1-octadecanethiol. All of the atoms were treated as chemicallyactive atoms in this truncated model. This scheme is designed tominimize computational cost while allocating additional basisfunctions for describing parts of the system that undergo signifi-

(53) Ardalan, P.; Bent, S. F. Unpublished results.(54) Tillman, N.; Ulman, A.; Schildkraut, J. S.; Penner, T. L. J. Am. Chem. Soc.

1988, 110, 6136–6144.(55) Snyder, R. G.; Strauss, H. L.; Elliger, C. A. J. Phys. Chem. 1982, 86, 5145–

5150.

(56) Shirley, D. A. Phys. Rev. B 1972, 5, 4709–4714.(57) Heister, K.; Zharnikov, M.; Grunze, M.; Johansson, L. S. O.; Ulman, A.

Langmuir 2001, 17, 8–11.(58) Cavalleri, O.; Gonella, G.; Terreni, S.; Vignolo,M.; Pelori, P.; Floreano, L.;

Morgante, A.; Canepa, M.; Rolandi, R. J. Phys.: Condens. Matter 2004, 16,S2477–S2482.

(59) Miller, T.; Rosenwinkel, E.; Chiang, T.-C. Solid State Commun. 1983, 47,935–938.

(60) Ranke, W.; Jacobi, K. Surf. Sci. 1977, 63, 33–44.(61) Mui, C.; Han, J. H.;Wang,G. T.;Musgrave, C. B.; Bent, S. F. J. Am. Chem.

Soc. 2002, 124, 4027.(62) Widjaja, Y.; B, M. C. Surf. Sci. 2000, 469, 9.(63) Widjaja, Y.; M, M. M.; B, M. C. J. Phys. Chem. B 2000, 104, 2527.(64) Trucks, G. W.; Raghavachari, K.; Higashi, G. S.; Chabal, Y. J. Phys. Rev.

Lett. 1990, 65, 504.(65) Hohenberg, H.; Kohn, W. Phys. Rev. B 1964, 1, 36B864.(66) Kohn, W.; Sham, L. Phys. Rev. A 1965, 140, 1133.(67) Becke, A. D. Phys. Rev. A 1988, 38, 3098.(68) Lee, C. T.; Yang, W. T.; Parr, R. G. Phys. Rev. B 1988, 37, 785.

8422 DOI: 10.1021/la904864c Langmuir 2010, 26(11), 8419–8429

Article Ardalan et al.

cant modification to the electronic structure during reaction. It isimportant to point out that there are two important time scalesassociated with the self-assembly process, namely, (1) chemisorp-tion of the headgroup on the substrate which typically occurs onthe time scale of milliseconds to minutes,46 and (2) alkyl chainreorganization (on the order of hours to days),46 which is thekinetically limiting step for SAM formation. The truncatedmodel(1-ethanethiol) employed in these calculations serves to addressonly the first step, chemisorption.

A frequency calculation was performed after each geometryoptimization to determine zero-point energies and to verify thatminima and transition states have zero and only one imaginaryfrequency, respectively. Intrinsic reaction coordinate (IRC) com-putations were carried out for key transition states to check if theyconnect the desired minima.Moreover, all of the transition stateswere visually inspected to ensure that the imaginarymodes corres-ponded to the correct reactions. All optimization and frequencycalculations were completed in the gas phase.

The single-point energies were calculated using a more exten-sive mixed basis set by applying to the gas-phase-optimized struc-tures in IPA media, and energies from the higher level of theoryare reported. For this, the triple-ζ 6-311þþG (d, p) basis set wasemployed, except for the seven subsurface germanium atoms,which were described using the LANL2DZ effective core poten-tial and basis. This approach has been found to reproduce experi-ments and the results of high level methods such as quadratic con-figuration interaction singles and doubles and connected triples(QCISD(T)) relatively accurately.69 Solvent effects were modeledusing the conductor-like polarizable continuum model (CPCM),70,71 and the atomic radii from the universal force field (UFF)were used for the solute atomic radii. The CPCM solvent para-meters were selected for IPAwith a dielectric constant of 18.3,72 acalculated solvent probe radius of 3.69 A, and a solvent density of0.0079 molecule/A3.72 All relative energies reported herein arerelative energies in solution phase correctedwith zero-point energiesobtained from the calculated gas phase frequencies. All calcula-tions were performed using the Gaussian 03 software package.73

3. Results and Discussion

3.1. Density Functional Theory.Webeginwith a discussionof DFT results. In this study, we explored several different possi-ble reactions between 1-alkanethiols and halogen-terminatedGe(100) surfaces. Figure 1 illustrates these reactions with mono-halides atGe(100) surfaces, namely,HX (X=Cl, Br) elimination,HX elimination followed by insertion, and dimer bond cleavage.

Figure 2 shows the calculatedHX (X=Cl, Br) elimination andHX elimination/insertion pathways for the reaction of 1-ethane-thiol with Cl- and Br-terminated Ge(100) surfaces. In Figure 2,the solid lines represent two consecutive elimination pathways,

while the dashed lines represent the first elimination pathwayfollowed by the insertion pathway. Moreover, pathways whichare shown with red and black colors correspond to 1-ethanethiolreactions at Br- and Cl-terminated Ge(100), respectively. It isevident that, for all of the reactions studied, energies are above theentrance channel, which makes each of these reactions thermo-dynamically unfavorable. However, a constant flux of reactantsand removal of the hydrohalogenic acid product may shift thesereactions toward completion even at room temperature. The endo-thermicity of the surface elimination reactions can be explained interms of the bond strength differences between the bonds thatform (Ge-S and H-X) and the bonds that break (Ge-X andS-H) during the reaction. For example, for the HCl eliminationreaction, the total energy of the Ge-Cl bond (∼88 kcal/mol) andS-H bond (∼80 kcal/mol) broken in the reaction is higher thanthe total energy of the Ge-S bond (∼60 kcal/mol) and H-Clbond (∼103 kcal/mol) formed. The energetic cost to undergo theHCl elimination reaction estimated by the bond energy calcula-tion (∼5 kcal/mol) is very close to the endoenergicity shown in thepathway in Figure 2 (4.3 kcal/mol).

Although the general mechanism of the hydrohalogenic acidelimination does not depend on the surface termination, thereaction barriers and energetics of the intermediates and productsshow differences based on the identity of surface halides. Forreactions at Cl- and Br-terminated Ge(100) surfaces, the firsttransition states for HCl and HBr elimination pathways are 20.1and 21.9 kcal/mol above the entrance channel, respectively. Thisleads to a hydrogen-bonded intermediate before eliminating HClorHBr and forming a surfacemonothiolate at 6.2 and 9.6 kcal/molabove the entrance channel on Cl- and Br-terminated Ge(100)

Figure 1. Possible surface reaction products of 1-alkanethiols andmonohalides at Ge(100) surfaces.

Figure 2. Reactions of 1-ethanethiol on Cl- and Br-terminatedGe(100); HCl or HBr elimination and insertion pathways.

(69) Wang, G. T.; Mui, C.; Tannaci, J. F.; Filler, M. A.; B, M. C.; F, B. S.J. Phys. Chem. B 2003, 107, 12256.(70) Barone, V.; Cossi, M. J. Phys. Chem. A 1998, 102, 1995.(71) Cossi, M.; Rega, N.; Scalmani, G.; Barone, V. J. Comput. Chem. 2003, 24,

669–681.(72) CRC Handbook of Chemistry and Physics, 81st ed.; CRC Press LLC: Boca

Raton, FL, 2000-2001.(73) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.;

Cheeseman, J. R.; Montgomery, J. J. A.; Vreven, T.; Kudin, K. N.; Burant, J. C.;Millam, J. M.; Iyengar, S. S.; Tomasi, J.; Barone, V.; Mennucci, B.; Cossi, M.;Scalmani, G.; Rega, N.; Petersson, G. A.; Nakatsuji, H.; Hada, M.; Ehara, M.;Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.;Kitao, O.; Nakai, H.; Klene,M.; Li, X.; Knox, J. E.; Hratchian, H. P.; Cross, J. B.;Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin,A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Ayala, P. Y.; Morokuma, K.;Voth, G. A.; Salvador, P.; Danenberg, J. J.; Zakrzewski, V. G.; Dapprich, S.;Daniels, A. D.; Strain, M. C.; Farkas, O.; Malick, D. K.; Rabuck, A. D.;Raghavachari, K.; Foresman, J. B.; Ortiz, J. V.; Cui, Q.; Baboul, A. G.; Clifford,S.; Cioslowski, J.; Sfefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi,I.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Peng, C. Y.;Nanayakkara, A.; Challacombe, M.; Gill, P. M. W.; Johnson, B.; W., C.; Wong,M.W.; Andres, J. L.; Gonzalez, C.; Pople, J. A.,Gaussian 03, revision B.3; Gaussian,Inc.: Pittsburgh, PA, 2003.

DOI: 10.1021/la904864c 8423Langmuir 2010, 26(11), 8419–8429

Ardalan et al. Article

surfaces, respectively. After this step, a second 1-ethanethiol mole-cule can react with the remaining surface halides with a similarreactionmechanism as the first elimination step, but with a slightlyhigher barrier from the monothiolate intermediate. Based onsimple kinetics calculations, the reaction of the first thiolmoleculewith the unreacted halide surfaces is 10-100 times faster than thereaction on the monothiolated dimers (i.e., second eliminationstep). In addition, due to smaller barriers for the reverse reaction,desorption of monothiolates may compete with formation of thesecond thiolate, although it requires the presence of nearby HClmolecules. Because the halogen is completely removed only ifthiolates form at both Ge atoms of the dimer, this suggests thatsome residual chlorine or bromine may remain at the surface, aswill be confirmed by the SR-PES results (section 3.2). Althoughnot calculated directly, hydrohalogenic acid elimination reactionssimilar to those calculated for 1-ethanethiol reactions at halo-genated Ge(100) are expected to be kinetically favorable athalogenated Ge(111) surfaces.

Figure 2 also shows that the insertion of the sulfur, which entailsdimer bond cleavage and reconfiguration of the surface adducts,requires surmounting barriers 46.1 and 46.9 kcal/mol above themonothiolated state at Cl- and Br-terminated Ge(100) surfaces,respectively. Overall, these transition states are >50 kcal/molabove the entrance channel, which makes the insertion pathwayskinetically unfavorable at room temperature.

The kinetics of the alkanethiol attachment reaction (HX elimi-nation pathway) is expected to depend upon the halogen atominvolved, due to a variety of factors, including differences in electro-negativity, atomic size, and bond strength. Each of these factorsfavors the reaction on the Cl-terminated surface, consistent withthe DFT results in Figure 2, which shows more favorable ener-getics for each of the Cl reactions compared to the Br reactions.For example, the Pauling electronegativity difference between Geand Br (0.95)72 is less than that between Ge and Cl (1.15),72 andthus, the Ge backbonds are more polarized when attached to Clatoms than to Br atoms. This bond polarization is expected tohelp lower the activation barrier for the proposedHX eliminationreactions, leading to a lower activation barrier in the case ofchlorine. In addition, the larger size of the Br atommay affect thekinetics by introducing larger steric hindrance in the reaction thanfor Cl atoms. Also, bromine forms weaker bonds compared tochlorine in general which affects the overall energetics. Hence, acombination of sterics and bond polarization leads to kineticallyand thermodynamically more favorable 1-alkanethiolate chemi-sorption at Cl-terminated Ge surfaces. Although these calcula-tions shed light on the bonding of the headgroup to the surface,the second step of self-assembled monolayer formation, that is,alkyl chain reorganization, is not captured here.

The dimer bond cleavage pathway illustrated in Figure 1requires Ge-Ge bond scission with transfer of H from the thiolmolecule to one of theGedimer atoms.The calculatedbarriers forthis pathway on both surfaces is high (∼30 kcal/mol, data notshown), suggesting that it is not kinetically competitive at roomtemperature and should not contribute to the surface productsafter thiolation. We note that the energetics of the dimer bondcleavage pathway are nearly identical for reactions at the Cl- andBr-terminated surfaces. This similarity in energetics can beattributed to the fact that these reactions leave the surface halidespecies intact.

The kinetics and thermodynamics of the self-assembly processbecome complicated when considering various solvent effects,and they are not very well understood.46 We have previouslyfound experimentally that using polar protic solvents such as IPAresults in well-packed 1-octadecanethiolate SAMs at halogenated

Ge surfaces, whereas we were unable to generate good qualitySAMs of the same molecules using nonpolar solvents such ashexane or toluene.36 Although some references suggest fasterSAM formation after using nonpolar solvents such as hexane,other studies have shown that these SAMs can be less organizeddue to strong adsorbate-solvent interactions.46 Solvents mayalso interact with the substrate and possibly adsorb at the surfaceinitially, followed by their displacement by the SAMmolecules.74

Nevertheless, all of these interactions are difficult to model indetail. Consequently, in the current work, the solvent effect iscaptured implicitly by employing the CPCM model. An evenlarger energy stabilization is expected when surface adducts areallowed to interact explicitly with the solvent molecules; however,the approach employed in this work is still useful to predict themost favorable mechanism for 1-alkanethiol adsorption at thehalogenated Ge(100) surfaces.

The theoretical results represented in this section are intendedto provide guidance for experimental studies of alkanethioladsorption on germanium surfaces. The DFT calculations yieldseveral key predictions: (1) residual Cl and Br will be left at thesurface after the thiolation; (2) the initial thiolation reaction(attachment of the headgroup) will be faster on the Cl-terminatedGe surfaces than on the Br-terminated Ge surfaces; (3) dimerbond cleavage reactions (via insertion) will not be observed to anysignificant extent, and hence, theGe-Gedimer bondswill remainintact; (4) higher temperatures will favor the formation of thiolateproducts, since the reactions are endoenergetic. The experimentalresults described below will offer direct confirmation for the firstprediction, that is, the presence of residual halide at the surface.The other predictions will require future study for substantiation.3.2. SR-PES. 3.2.1. Bonding at Room Temperature.

Figure 3 shows the Ge(3d), S(2s), and C(1s) core level spectrataken at room temperature after the Cl-terminated Ge(111), Cl-terminated Ge(100), and Br-terminated Ge(100) samples weredipped in the ODT solution for 72 h. These samples are labeled inthe figure as “Ge(111)-Cl þ ODT”, “Ge(100)-Cl þ ODT”, and“Ge(100)-BrþODT”, respectively. The spectra corresponding toeach SAM-coated surface are complex and can be fit to severalpeaks. Table 1 lists the positions and relative intensities of thefitted peaks in the Ge(3d) core level spectra from Figure 3a andthe species to which they are assigned.We will discuss these peaksin detail below.

For the thiolated Ge(111) surface, five components are used tofit the Ge(3d) data in Figure 3a. Apart from theGe(3d) bulk peak(Ge0þ), four other contributionswith chemical shifts of 0.33, 0.72,1.56, and 2.49 eV are also observed (see Table 1). The latter twoshifts correspond well to peaks reported in the literature for Gebonded to O with oxidation states of þ2 (Ge2þ(O)) and þ3(Ge3þ(O)), respectively23,75 The total oxide to bulk ratio is small(∼0.06), and hence, the Ge oxide species are minority adductsat the Ge(100) surface. The presence of a small amount ofGe oxide is further confirmed by the O(1s) core level peaks (videinfra).

The analysis of the 0.33 and 0.72 eV shifted peaks is moredifficult due to limitations in energy resolution at the 200 eVphoton energy (∼0.3 eV). The 0.33 eV shift corresponds well withGe1þ attached to S in the form of Ge monothiolate (Ge1þ(S)).In fact, Weser et al.14 and Roche et al.13 have reported 0.33 and0.4 eV chemical shifts to the Ge(3d) peak per Ge-S bond, res-pectively. Based on this argument, a 0.66-0.8 eV shift attributed

(74) Himmelhause, M.; Gauss, I.; Buck, M.; Elisert, F.; Woll, C.; Grunze, M.J. Electron Specrosc. Relat. Phenom. 1998, 92, 139–149.

(75) Adhikari, H.; McIntyre, P. C.; Shiyu, S.; Pianetta, P.; Chidsey, C. E. D.Appl. Phys. Lett. 2005, 87, 263109.

8424 DOI: 10.1021/la904864c Langmuir 2010, 26(11), 8419–8429

Article Ardalan et al.

to Ge2þ(S) species (Ge dithiolate) can also be expected if therewere such sites available at theGe(111) surface after chlorination.However, no dichloride sites were observed on the initially Cl-terminatedGe(111) surface.23,36 Complicating the assignment, wehave also observed residual chlorine bonded to the surface after

thiolation (see the Cl(2p) core level spectra supplied in theSupporting Information). From that photoemission data, thecoverage of Cl at the Ge(111) surface is estimated as ∼0.2 MLafter thiolation. A 0.6 eV Ge(3d) shift has been reported for ∼1MLGemonochloride (Ge1þ(Cl)) generated onGe(111) afterHCl

Figure 3. Room temperature (a) Ge(3d) core level spectra taken at incident photon energy of 200 eV, (b) S(2p) core level spectra taken at300 eV, and (c) C(1s) core level spectra taken at 350 eV after various halogenated Ge surfaces were dipped in ODT solution for 72 h. All thecurves in (a) are normalized to the height of the bulk peak to emphasize the peak shape difference. All the curves in (b) and (c) are normalizedby the incident synchrotron radiation beam flux to stress the peak intensity difference.

DOI: 10.1021/la904864c 8425Langmuir 2010, 26(11), 8419–8429

Ardalan et al. Article

treatment.23,36 Hence, the 0.72 eV shifted peak observed inFigure 3a after thiolation may stem from Ge monochlorideand/or Ge dithiolate. The presence of dithiolate at Ge(111)surface would be interesting, since it suggests that thiolationcan change the Ge(111) surface structures such that there aresites at the surface which accommodate two sulfur atoms per Geatom.

The presence of Ge-S bonds is further confirmed by analysisof the S(2p) peak shown in Figure 3b. This peak can be fit to threecomponents. These peaks include one from surface thiolates,which contributes to the majority of the S peak as has beenreported on metals46,57,58,76-79 and other semiconductors such asGaAs(001),49,51 as well as two peaks with chemical shifts of-0.95andþ0.60 eV in kinetic energy (KE). The peak shifted by-0.95 eVcorresponds to unbound thiols.49,57,78,79 The presence of unboundthiol at room temperature indicates presence of unreacted physi-sorbedODTmolecules (∼ 21%of the total S peak). Furthermore,our results do not show evidence of sulfonates (-SO3

-) orsulfinates (-SO2

-) at lower kinetic energies, which indicates thatthe thiolates have not been oxidized.46,74,80,81

The peak at þ0.60 eV in KE is assigned to atomic S or otherforms of S.58,78,79,82 This “atomic sulfur” peak, which herecontributes to only∼5% of the total S peak, has been previouslyreported for alkanethiol/Au(111) systems. However, clear assign-ment of this peak has been the subject of debate in the literature,and its presence appears todependon the experimental conditionsand the chain length of the thiol molecule. In such systems, thispeak has been observed during the initial stages of SAM growth,

in a low coverage SAM after annealing in air, and after vacuumannealing at higher temperatures.77,79 At room temperature, C-Sbond scission and formation of sulfide are quite unlikely. Never-theless, observation of this peak at room temperature has alsobeen attributed to dilute atomic sulfur or other thiol moleculesthat do not undergomolecular decomposition (e.g., sp hybridizedsulfur).77,79,83

Figure 3c shows that the C(1s) core level peak is asymmetric.Besides the alkyl carbon, a peak shifted 0.80 eV toward lower KEis necessary for good peak fitting. We attribute this small peak toC-S species based on the literature.74,84-86 Using this assign-ment, the ratio of the C-S carbon to the alkyl C is 1/21, which isslightly lower than the 1/17 value expected for a C18H37S-Gespecies, suggesting that part of the bulk C signal may arise fromadventitious materials.

Although the general trends are the same, some differences areobserved between the Ge(100) and Ge(111) surfaces. Analysisof the Ge(3d) peak after thiolation of halogenated Ge(100)(Figure 3a) reveals differences from that of Ge(111). In the caseof ODT SAM onCl-terminated Ge(100), the spectrum shows thepresence of five peaks shifted to lower KE with respect to thebulk Ge peak. These peaks are shifted by 0.33, 0.74, 1.24, 1.59,and 2.49 eV, and based on the previous arguments can be attri-buted to Ge monothiolate, Ge dithiolate and/or Ge monochlo-ride, Ge dichloride (Ge2þ(Cl)),22,36 GeO, and Ge2O3, respectively(also see Table 1). As will be shown in the next sections, residualchemically bonded chlorine is present at the Ge(100) surface evenafter thiolation. In contrast to halogenation of Ge(111), previousstudies have revealed the presence of di- and monohalidesat Ge(100) after halogenation, with a higher concentration ofdihalides compared to monohalides; thus, such sites are availablefor formation of dithiolates on Ge(100) after ODT dipping.Although, due to the limitation on the energy resolution, the0.74 eV-shifted peak cannot be fully resolved; this peak may havecontributions from both Ge monochloride and Ge dithiolate.

Examination of the S(2p) peak (Figure 3b) shows the presenceof similar sulfur species at the thiolated Ge(100):Cl surface as ontheGe(111):Cl surfacewith the following key differences.Overall,the total S and thiolate integrated S(2p) peak area is ∼2 timeslarger on the Ge(100) surface. In addition, the atomic sulfurcontributes a larger fraction (∼12%) of the total S(2p) area. Thelarger thiolate signal suggests that, for the same dipping time,more thiols react at the Ge(100) surface compared to the Ge(111)surface. In fact, we have already reported that ODT SAMs for-med at Ge(100) surfaces have higher water contact angles,thicknesses, and ambient stability than those formed at Ge(111)surfaces.36 The current results are consistent with those observa-tions. This is also apparent inFigure 3cwhere the total C coverageis ∼1.4 times higher on Ge(100) compared to Ge(111) after thio-lation. Observation of more Cl remaining after thiolation at the Ge-(111):Cl surface (∼0.2 ML on Ge(111):Cl vs∼0.1 ML on Ge(100):Cl; Supporting Information, Figures S2 and S3) further supportsthis argument (i.e., less Cl reacts at the Ge(111) after thiolation).

Investigation of the ODT coatedGe(100):Br surface (Figure 3a)reveals a similar shapedGe(3d) peak as for the thiolatedGe(100):Cl surface. The peak can be fit to a bulk Ge peak as well as five

Table 1. Positions and Relative Intensities of the Fitted Peaks in the

Ge(3d) Core Level Spectra from Figure 3a and the Species to

Which They Are Assigned

specieskinetic energy

(eV)intensity relative to

the bulk Ge

Ge(lll)-Cl þ ODT

bulk Ge 166.28 1.00Ge1þ(S) 165.95 0.35Ge1þ(C1) þ Ge2þ(S) 165.56 0.22Ge2þ(O) 164.72 0.05Ge3þ(O) 163.79 0.07

Ge(100)-Cl þ ODT

bulk Ge 166.28 1.00Ge1þ(S) 165.95 0.40Ge1þ(C1) þ Ge2þ(S) 165.56 0.17Ge2þ(C1) 165.04 0.12Ge2þ(O) 164.69 0.10Ge3þ(O) 163.79 0.08

Ge(100)-Br þ ODT

bulk Ge 166.28 1.00Ge1þ(S) 165.93 0.40Ge1þ(Br) þ Ge2þ(S) 165.58 0.16Ge2þ(Br) 165.18 0.09Ge2þ(O) 164.73 0.11Ge3þ(O) 163.82 0.09

(76) Yang, Y. W.; Fan, L. J. Langmuir 2002, 18, 1157–1164.(77) Ishida, T.; Choi, N.;Mizutani,W.; Tokumoto, H.; Kojima, I.; Azehara, H.;

Hokari, H.; Akiba, U.; Fujihira, M. Langmuir 1999, 15, 6799–6806.(78) Vericat, C.; Vela, M. E.; Andreasen, G.; Salvarezza, R. C. Langmuir 2001,

17, 4919–4924.(79) Vericat, C.; Vela, M. E.; Benitez, G. A.; Martin Gago, J. A.; Torrelles, X.;

Salvarezza, R. C. J. Phys.: Condens. Matter 2006, 18, R867–R900.(80) Whelan, C. M.; Malcolm, R. S.; Barnes, C. J. Langmuir 1999, 15, 116–126.(81) Willey, T. M.; Vance, A. L.; van Buuren, T.; Bostedt, C.; Terminello, L. J.;

Fadley, C. S. Surf. Sci. 2005, 576, 188–196.(82) Chang, Z.; Tang, W. H. Surf. Sci. 2007, 2005–2011.

(83) Ishida, T.; Hara,M.; Kojima, I.; Tsuneda, S.; Nakoi, N.; Sasabe, H.; Knoll,W. Langmuir 1998, 14, 2092–2096.

(84) Brito, R.; Tremont, R.; Feliciano, O.; Cabrera, C. R. J. Electroanal. Chem.2003, 540, 53–59.

(85) Petrovykh, D. Y.; Kimura-Suda, H.; Opdahl, A.; Richter, L. J.; Tarlov,M. J.; Whitman, L. J. Langmuir 2006, 22, 2578–2587.

(86) Moulder, J. F.; Stickle, W. F.; Sobol, P. E.; Bomben, K. D. Handbook ofX-ray Photoelectron Spectroscopy; Perkin-Elmer Corp. Physical Electronics Division:Eden Prairie, MN, 1992.

8426 DOI: 10.1021/la904864c Langmuir 2010, 26(11), 8419–8429

Article Ardalan et al.

additional peaks shifted by 0.35, 0.70, 1.10, 1.55, and 2.46 eVwithrespect to bulk Ge (see Table 1). These shifts can be attributed toGe monothiolate, Ge dithiolate and/or Ge monobromide(Ge1þ(Br)), Ge dibromide (Ge2þ(Br)),36,87 GeO, and Ge2O3,respectively. This is similar to ODT coated Ge(100):Cl, thoughthe shifts due to residual bromine (∼0.3 ML is present afterthiolation; see the Supporting Information, Figure S2) are ex-pected to be smaller due to the lower electronegativity value forBr compared to Cl. Moreover, examination of the S(2p) feature(Figure 3b) for the thiolated Ge(100):Br surface shows thepresence of similar species to that of thiolated Ge(100):Cl. Thetotal integrated peak area for S(2p) is similar (only ∼1.1 timeshigher) to that on Ge(100):Cl. However, the ratio of unboundthiol to thiolate is ∼1.7 times lower on Ge(100):Br than onGe(100):Cl. Also, the total amount of surface thiolates is similaratGe(100):Cl andGe(100):Br. TheCphotoemission data supportthe conclusion that a similar thiolated coverage is observed onGe(100):Br as on Ge(100):Cl, with the data (Figure 3c) showing∼1.2 times more total C on thiolated Ge(100):Cl compared tothiolated Ge(100):Br.

Another interesting observation is that the C peak for thethiolated Ge(111):Cl is shifted by ∼0.3 eV to the higher KE side,with respect to the same peak that corresponds to thiolatedGe(100):Cl spectrum (Figure 3c). A peak shift, albeit smaller, isalso seen for the thiolatedGe(100):Br surface. This effect has beenpreviously attributed to differences in the SAM thickness88 (i.e.,thicker SAMs are better insulators, and hence, they can betterdischarge the positive charge generated by the photoelectronemission). In fact, our ellipsometry measurements show that thethickness values of theODT films formed atGe(100):Cl,Ge(111):Cl, and Ge(100):Br surfaces are 18.6( 0.5, 8.3 ( 1.2, and 13.9(0.9 A, respectively; hence, those shifts in the C(1s) peaks areconsistent with the expected effect.

Both the Ge(3d) photoemission data and the O(1s) photo-emission data show the presenceof a small amountofGeoxides atODT coated Ge(100) and Ge(111) surfaces formed from thehalides. Although part of this can be attributed to Ge oxidationduring SAM formation, we suspect that ambient exposure of thesurfaces before characterization is another source of Ge oxida-tion. Nevertheless, both Ge halides and oxides are only minorityadducts at the thiolated Ge surfaces. On the whole, as depicted inFigure 4, our data show that Ge monothiolate and possibly Gedithiolate make up the majority of chemically bonded species onboth Ge(111) and Ge(100) surfaces at room temperature.

To summarize, the PES data indicate the following: afterthiolation of halide-terminated germanium, both the Ge(100)and (111) surfaces contain monothiolates and possibly dithio-lates. In addition, unbound thiol and atomic sulfur are also obser-ved at the surface, together with residual halide. The observationof residual halide is consistent with theDFT results, which predictthat the elimination reaction will not proceed to completion,leaving some halogen unreacted at the surface.

3.2.2. Thermal Stability. A series of studies was carried outto investigate the behavior of the thiolate SAMs upon thermalannealing. Figures 5-8 depict the results of photoemissionmeasurements after stepwise vacuum annealing of the ODTSAMs formed on Cl-terminated Ge(100) (Ge(3d), S(2p), andC(1s) core level spectra of theODTcoatedBr-terminatedGe(100)and Cl-terminated Ge(111) surfaces after vacuum annealing to310 �C are shown in the Supporting Information, Figures S2 and

S3). The following analysis of the temperature-dependent dataleads to these conclusions about the thermal stability and decom-position pathways of the thiolate SAMs at the Ge surface: (1) thethiolate SAMs are stable to nearly 150 �C; (2) decomposition ofthe SAMs at temperatures between 180 and 350 �C leads toformation of surface carbide and sulfide; (3) above 430 �C, thesurface sulfide is completely removed but the carbide, plus a smallamount of oxygen bound to carbon remains; (4) residual chlorineremaining on the surface after thiolation of the Cl-terminatedsurface desorbs completely by 280 �C.

The data in Figures 5-7 indicate that the majority of the SAMmolecules are stable to at least 150 �C. Changes in the Ge(3d)peak, as well as in the C(1s) and S(2p) peaks, are relatively minorup until this temperature. In addition, the detailed evolution ofthe S(2p) peak upon vacuum annealing (Figure 6) indicates thatthe thiolate species are stable until above 150 �C. According toFigure 6, the unbound thiol peak decreases monotonically up to180 �C, before completely disappearing upon 280 �C anneal. Atthe same time, the component that corresponds to Ge thiolatesincreases by a factor of∼1.4 after annealing to 150 �C. By 150 �C,the ratio of the unbound thiol to thiolates drops to 73% of that atroom temperature. Consequently, we propose that the increase inthe thiolate signal can be partly related to conversion of unboundthiol species to surface thiolates as the temperature is raised. Ateven higher temperatures (above 150 �C), the intensity of thethiolate S(2p) peak begins to decrease, but it is not completelyremoved until 430 �C when all other S components desorb fromthe Ge(100) surface. Consequently, these results show that Gethiolates are stable at the Cl-terminated Ge(100) surface up to atleast 150 �C.

Minor changes can be seen in other SR-PES spectra at tempe-ratures below 180 �C. For example, Figure 7 shows that the C(1s)peak at 61.1 eV corresponding to the alkyl chains (-C-C-) isbroadened and shifted to higherKE after vacuumannealing. Thispeak broadening and shifting has been previously attributed tostructural changes in the alkyl chains and to an electrostaticscreening effect.83 Park et al. pointed out that the shift in the Cpeak is due to changes in the polarizability of the SAMs afterstructural changes.89 After the SAMs are annealed to highertemperatures, they become poorer insulators and therefore, thedamaged SAMs experience less charging compared to pristineSAMs.83 Overall, this peak is shifted by ∼0.3 eV to higher KE at280 �C compared to room temperature.

More significant spectral changes are observed beginning attemperatures of 280 �C and higher. As will be described below,these changes indicate decomposition of the SAMs with conco-mitant formation of surface sulfides and carbides. According toFigure 5, the ratio of the S- as well as O-induced components of

Figure 4. Molecularmodels representingGemonothiolate (Ge1þ(S))and Ge dithiolate (Ge2þ(S)) moieties after room temperature thio-lation of halogenated Ge(100) and Ge(111) surfaces.

(87) Sun, S. Germanium Surface Cleaning, Passivation, and Initial Oxidation.Ph.D. Thesis, Stanford University, 2007.(88) Ishida, T.; Nishida,N.; Tsuneda, S.; Hara,M.; Sasabe,H.; Knoll,W. Jpn. J.

Appl. Phys. 1996, 35, L1710–L1713.(89) Park, J.-S.; Nguyen, A.; Barriet, D.; Shon, Y.-S.; Randall Lee, T. Langmuir

2005, 21, 2902–2911.

DOI: 10.1021/la904864c 8427Langmuir 2010, 26(11), 8419–8429

Ardalan et al. Article

the Ge(3d) peak relative to bulk Ge decreases, and this occurs inconjunction with a significant decrease in the S(2p) thiolate andthe O(1s) Ge oxide peaks (Figure 6; oxygen data are compiled inSupporting Information, Figure S1). A peak with -0.32 eV shiftis the only feature observed on the lower KE side after annealingto 350 �C; this can be attributed to Ge with an oxidation state of1 attached to S (i.e., Ge thiolate and/or Ge sufide). After 430 �Cvacuum annealing, peak fitting shows the presence of a smallfeature at lower KE, which is-0.37 eV shifted with respect to thebulk Ge value. Appearance of this peak is correlated with newpeaks in the C(1s) (Figure 7) spectrum, assigned to surfacecarbide, as will be discussed below. Vilcarromero and Marqueshave reported various peak shifts in the Ge(3d) orbital as a result

of alloying with C.90 Based on that report and the change in theC(1s) spectrum, we attribute this peak shift to Ge1þ(C) (Gecarbide). This Ge(3d) peak associated with Ge carbide remainsstable at the surface after vacuum annealing to above 470 �C.

Another interesting change in the Ge(3d) spectrum is theappearance of two new components on the higher KE side ofthe bulk peak after vacuum annealing, appearing at temperaturesas low as 60 �C. One has a chemical shift of ∼þ0.3 eV and theother has a chemical shift ofþ0.54 eV. Clear assignment of thesesurface states has been the subject of some debate in the literature;however, both of these components are generally associated withsurface state peaks andgrowwith the annealing temperature.44,87,91

Figure 7 shows major changes in the C(1s) SR-PES peak attemperatures above 280 �C, with a third peak (shifted byþ0.7 eVin KE) becoming dominant by 430 �C. The data reveal growth of

Figure 5. Ge(3d) core level spectra taken at 200 eV photon energyfrom a thiolated Ge(100):Cl surface after step by step vacuumannealing. The substrate was prepared by dipping Cl-terminatedGe(100) surface in ODT solution for 72 h. All the curves arenormalized to the height of the bulk peak to emphasize the peakshape difference.

Figure 6. S(2p) core level spectra taken at 300 eV photon energyfrom a thiolated Ge(100):Cl surface after step by step vacuumannealing. The substrate was prepared by dipping Cl-terminatedGe(100) surface in ODT solution for 72 h. All the curves arenormalized by the incident synchrotron radiation beam flux tostress the peak intensity difference.

(90) Vilcarromero, J.; Marques, F. C. Appl. Phys. A: Mater. Sci. Process. 2000,70, 581–585.

(91) Cao, R.; Yang, X.; Terry, J. Phys. Rev. B 1992, 45, 13749–13752.

8428 DOI: 10.1021/la904864c Langmuir 2010, 26(11), 8419–8429

Article Ardalan et al.

this third peak beginning as low as 150 �C. This peak can beattributed to Ge carbide arising from SAM decomposition, inagreement with the literature.90 In fact, Ishida et al. reported a0.8 eV shift for this peak with respect to the alkyl C peak afterODTSAMonAu(111) was annealed to 200 �Cunder vacuum for1 h, and they attributed this new peak to SAM decomposition aswell as formation of a so-called “striped phase” in which ODTmolecules lie mostly parallel to the surface.83 Based on thisobservation, the -0.32 eV shift observed in the Ge(3d) spectrumafter 350 �C vacuum annealing (Figure 5) can be assigned at leastpartially to Ge carbides, although we cannot distinguish the Gecarbide andGe sulfide peaks due to energy resolution limitations.

In the C(1s) spectrum, carbide becomes the dominant feature after430 �Cand this peak is stable even after annealing to above 470 �C.The spectrum also shows the presence of a new C feature (-1.3 eVin KE) which contributes to only ∼2% of the total C and can beattributed to oxidized C.86 Finally, Figure 7 also shows that the(-C-S-) peak disappears after annealing to 430 �C, which isconsistent with removal of S from the Ge(100) surface (Figure 6).

Finally, examination of the S(2p) peak upon annealing to highertemperatures reveals significant growth of a third component athigher KE, which cannot be attributed to either unbound thiol orthiolate.Rather, this peak is assigned to surface sulfide.AhigherKE(lower binding energy) peak has also been previously reportedwhenfully covered ODT SAMs on Au(111) were annealed to 200 �Cunder vacuum for 1 h, and assigned to atomic sulfur as a result ofC-S cleavage.83 We believe that after annealing to above roomtemperature this peak can be better assigned to Ge sulfide. In fact,this peak constitutes the majority of the S species on the Ge surfaceafter annealing to above 180 �C due to cleavage of the C-S bond.Annealing to 350 �C results in a decrease in the Ge sulfide S(2p)peak, followed by complete loss by 430 �C. This is in contrast to thesurface carbide, which is still present on the surface even at 430 �C.92

Figure 7. C(1s) core level spectra taken at 350 eV photon energyfrom a thiolated Ge(100):Cl surface after step by step vacuumannealing. The substrate was prepared by dipping Cl-terminatedGe(100) surface in ODT solution for 72 h. All the curves are normal-izedby the incident synchrotronradiationbeamflux to stress thepeakintensity difference.

Figure 8. Cl(2p) core level spectra taken at 300 eV photon energyfrom a thiolated Ge(100):Cl surface after step by step vacuumannealing. The substrate was prepared by dipping Cl-terminatedGe(100) surface in ODT solution for 72 h. All the curves arenormalized by the incident synchrotron radiation beam flux tostress the peak intensity difference.

(92) It is also important to note that irradiated induced damage to the SAMs hasbeen reported as the result of synchrotron radiation exposure. In fact, Ulman andcoworkers reported appearance and growth of þ1.2 eV in binding energy (-1.2 inKE) shifted peak (with respect to surface thiolates) after 1-alkanethiolate SAMs onAu(111) were exposed to synchrotron radiations and S(2p) core level wasmeasuredat ∼200 eV photon energy. ( Heister, K.; Zharnikov, M.; Grunze, M.; Johansson,L. S. O.; Ulman, A. Langmuir 2001, 17, 8–11.) Such a shift was attributed to dialkylsulfide moieties in the form of C 3 3 3 S-C. Such an effect is not expected to be the case inour study, since neither such peak shift value nor growth of such peaks was observed inthe S(2p) core level scans.

DOI: 10.1021/la904864c 8429Langmuir 2010, 26(11), 8419–8429

Ardalan et al. Article

The behavior of chlorine at theGe(100) surface as a function ofannealing temperature is also of interest. The Cl(2p) PES spectraof Figure 8 indicate that residual chlorine is clearly present on theCl-terminated Ge(100) surface even after 72 h exposure to ODTsolution and the formation of the thiolate SAM. This chlorineremainson the surfaceuntil above180 �C.After annealing to280 �C,the peaks corresponding to Ge chlorides disappear, in agreementwith desorption of the residual Cl from the Ge(100) surface.87

Investigation of the ODT coated Ge(100):Br surface aftervacuum annealing to 310 �C shows similar trends in the corelevel spectra as seen for ODT coated Ge(100):Cl. As evident inFigure S2 (Supporting Information), formation of Ge carbideand Ge sulfide is observed at higher temperatures, as well asdesorption of the bromine. Moreover, the studies of ODT coatedGe(111):Cl after annealing to 310 �C show similar trends, too.Core level spectra of this system upon annealing (SupportingInformation, Figure S3) reveal Ge sulfide and Ge carbide forma-tion and desorption of chlorine. However, we note the followingdifferences: Overall, the PES signal from Ge sulfide formed onGe(111) is lower than that on Ge(100), which is consistent withlower S reaction at the Ge(111) surface at any temperature(section 3.2.1). Moreover, the surface states in the Ge(3d) corelevel spectra of the SAM-covered Ge(111) surface have chemicalshifts of ∼0.3 and 0.65 eV to higher KE, compared to ∼0.3 and0.54 eV on Ge(100), which suggest a complex reconstructiondifferent at this surface than on Ge(100). However, clear assign-ment of these peaks has been debated.93

The results of our study reveal very good stability of the thiolateSAMS, up to temperatures as high as 150 �C. Maboudian and co-workers have studied the thermal stability of the ODT monolayerformed on H-terminated Ge(111) surfaces with high-resolutionelectron energy loss spectroscopy (HREELS).42 That study sug-gested no changes in the ODT monolayer after annealing to∼76 �C,whereas annealing to∼176 and∼276 �C resulted in partialand complete desorption ofODTmonolayer fromGe(111) surface,respectively. Moreover, no S was observed on the surface after∼276 �C annealing.42 Apart from uncertainty on the exact tempe-rature values reported herein ((20 �C), our results show clearevidence of Ge-S and S after vacuum annealing to above 310 �C.

4. Conclusions

Various characterization techniques together with DFT areemployed to study the reaction mechanism, surface bonding, andthermal stability of 1-octadecanethiolate self-assembled mono-layers (SAMs) at Cl- and Br- terminated Ge surfaces.

Our calculations show that hydrohalogenic acid eliminationreactions are kinetically favorable at these surfaces, with thermo-dynamically more favorable reactions occurring at Cl-terminatedGe surfaces. However, these reactions are endothermic, whichsuggests that removal of the products should facilitate thesereactions at room temperature. The calculations also show thatalternative reactions such as dimer cleavage or insertion arekinetically unfavorable and are not expected to be observed.Finally, the theoretical studies predict that residual halide con-centration may be present at the surface after thiolation due toincomplete reaction.

SR-PES results at room temperature indicate the presenceof residual Ge halide and Ge oxide at the surface after thio-lation. However, after ODT SAM formation on halogenatedGe(100) and Ge(111), both surfaces are mainly covered bymonothiolates and possibly dithiolates as well as unboundthiol and a small amount of atomic sulfur or other thiolswithout C-S bond cleavage. Higher levels of sulfur andcarbon are detected at the Ge(100) surface after thiolation,which indicates higher conversion of surface halides to surfacethiolates on Ge(100).

Vacuum annealing studies show that the Ge thiolates arethermally stable up to 150 �C. Furthermore, the majority of thesurface thiolates are converted to surface sulfide and carbide uponannealing to 350 �C on both Ge(100) and Ge(111) surfaces. Nosulfur is observed at the surface at 430 �C, whereas Ge carbide isstable to above 470 �C.

Acknowledgment. The authors are indebted to Dr. S. Sun,Dr. J.S.King,Dr.A.Paul,Dr.Z.Zhang,andDr.A.Mukhopadhyayfor insightful comments.We also thank the staff of the Center forPolymer Interfaces and Macromolecular Assemblies (CPIMA)and SSRL for their support. This work was supported by theNational Science Foundation [CHE 0615087 and CHE 0910717].Portions of this research were carried out at the Stanford Syn-chrotron Radiation Lightsource, a national user facility operatedby Stanford University on behalf of the U.S. Department ofEnergy, Office of Basic Energy Sciences.

Supporting Information Available: Cl and Br coveragecalculation at the thiolated Ge surfaces; O(1s) core levelspectra of a thiolated Ge(100):Cl surface after step by stepvacuum annealing; Ge(3d), S(2p), C(1s), Cl(2p), and Br(3d)spectra of thiolated Ge(100):Br and Ge(111):Cl surfaces atroom temperature and after 310 �C vacuum annealing. Thismaterial is available free of charge via the Internet at http://pubs.acs.org.

(93) Sieger,M. T.; Roesler, J.M.; Lin, D. S.;Miller, T.; Chiang, T.-C.Phys. Rev.Lett. 1994, 73, 3117–3120.