Molecule orientation in self-assembled monolayers determined
by infrared-visible sum-frequency generation spectroscopy
A.A. Mania, Z.D. Schultzb, B. Champagnec, C. Humberta, L. Dreesena,
A.A. Gewirthb, J.O. Whiteb, P.A. Thirya, A. Peremansa, Y. Caudanoa,?
aLaboratoire de Spectroscopie Mole ´culaire de Surface, Faculte ´s Universitaires Notre-Dame de la Paix,
Rue de Bruxelles, 61 B-5000 Namur, Belgium
bFrederick Seitz Materials Research Laboratory, University of Illinois at Urbana-Champaign,
104 S. Goodwin Ave., Urbana, IL 61801, USA
cLaboratoire de Chimie The ´orique Applique ´e, Faculte ´s Universitaires Notre-Dame de la Paix,
Rue de Bruxelles, 61 B-5000 Namur, Belgium
Available online 14 August 2004
The molecular orientation in self-assembled monolayers of thiophenol (TP) on Ag(1 1 1) and of para-nitroanilino-
dodecane-thiol (NAT) on Au was investigated by infrared-visible sum-frequency generation between 500 and 1500 cm?1.
resultsrevealthattheS–CbondofTPistilted37?fromtheAg(1 1 1)surfacenormal,withthearomaticringplaneperpendicular
to the surface. On Au, the tilt angle of the NAT molecule stabilizes below 60?at the end of film growth.
# 2004 Elsevier B.V. All rights reserved.
PACS: 68.43.Hn; 42.65.Ky; 68.43.Pq; 68.43.Fg
Keywords: Sum-frequency generation; Self-assembled monolayer; Adsorbate orientation; Vibration; Thiophenol; para-nitroanilino-dodecane-
Organic thin films and, in particular, self-
assembled monolayers (SAM) of conjugated mole-
cules are promising materials for building electronic,
optoelectronic, and photonic nanometer-scale devices
[1,2]. Many of these applications rely on the repro-
ducible manufacture of organized molecular arrays on
metallic substrates. Hence, knowing the molecule
orientation is vital and advocates the development
of investigation tools adapted to the determination
of interface order and adsorbate structure.
Infrared-visible sum-frequency generation (SFG)
spectroscopy makes use of two laser beams, one
Applied Surface Science 237 (2004) 444–449
? Corresponding author. Tel.: +32 81 725487;
fax: +32 81 724718.
E-mail address: email@example.com (Y. Caudano).
0169-4332/$ – see front matter # 2004 Elsevier B.V. All rights reserved.
visible and one tunable in the infrared, to record
interface-selective vibrational spectra. The intensity
of the directional SFG beam generated at the sum of
the incident frequencies increases resonantly whenthe
infrared beam excites an adsorbate vibration that is
both infrared and Raman active. The signal depends
on the relative orientations of the vibration infrared
and Raman-induced dipoles, with respect to the polar-
izations of the incident and SFG beams. These strict
selection rules lead to simple SFG signatures for SAM
on metals, that are highly sensitive to film order and to
adsorbate orientation and conformation.
By comparing ab initio calculations of the mono-
layer theoretical nonlinear optical response to experi-
mental SFG spectra acquired in the 500–1500 cm?1
range, we determined the orientation of the head
group in para-nitroanilino-dodecane-thiol SAM on
Au (NAT: p-NO2–C6H4–NH–(CH2)12–SH) and of
thiophenol on Ag(1 1 1) (TP: C6H5–SH). In order
to probe the out-of-plane deformations of the TP
carbon skeleton, the spectral range of our tunable
tabletop laser system was extended in the mid-IR
beyond the reach of conventional SFG set-ups (lim-
ited by the source to 900 cm?1unless a free-electron
laser is used [3–5]).
2. Experimental SFG set-up
The visible and tunable, infrared beams used for
SFG are produced by converting the 9400 cm?1radia-
tion of an all-solid-state Nd:YAG laser that is pumped
by flashlamps and mode locked by a frequency-dou-
bling nonlinear mirror combined with a two-photon
absorber (GaAs platelet) . The green visible beam
(18800 cm?1) is obtained by frequency doubling one
third of the YAG laser output in a BBO crystal. The
radiation in a picosecond optical parametric oscillator
In order to perform SFG spectroscopy as far as
500 cm?1, we used a KTP/CdSe OPO . The first
stage of the duo OPO relies on a KTP crystal opti-
mized for the 9400 cm?1! 5030 cm?1+ 4370 cm?1
(o!o þ e) SFG process. In the second step, 30 % of
the OPO idler beam (4175–4450 cm?1) is mixed with
the signal beam (5225–4950 cm?1) in a CdSe crystal
to produce, by difference-frequency generation, infra-
red radiation tunable from 1050 to 500 cm?1. The
powerat670 cm?1is2 mW(?0.8 mgJper8-pspulse).
For SFG spectroscopy, the p-polarized, co-propa-
gating IR and visible beams are synchronized and
focused on the sample with 65?and 55?incidence
angles to the surface normal, respectively. The p-
polarized SFG beam is detected by a photomultiplier
after spatial and spectral filtering using a Raman filter
and a double monochromator.
3. Theoretical SFG spectra of molecular layers
Extracting the precise orientation of adsorbed
molecules from IR-visible SFG spectra requires a
detailed knowledge of the molecule vibrational first
hyperpolarizability b. In the harmonic approximation,
far from electronic resonances, the contribution of
each vibration Qvis proportional to its Raman scatter-
ing and infrared absorption cross-sections :
vIR? vvþ iGv
where a and m are the molecular polarizability and
dipole moments. We can thus predict the SFG effi-
ciency from the computation of the vibration Raman
tensors and dynamic dipoles.
The adsorbed TP and NATorientations are defined
by the angles u, the molecule tilt with respect to the
surface normal, and j, the rotation of the benzene ring
around the molecular axis (so that, if u ¼ j ¼ 90?, the
molecules lie flat on the surface). If we assume a
uniform adsorption structure on the surface and an
isotropic molecule distribution among randomly
oriented domains, the SFG signal is evaluated to:
where Tðu;jÞ converts the molecular hyperpolariz-
ability to the laboratory frame, F holds the interface
Fresnel factors for ppp polarizations , NSis the
molecule density on the surface, and xð2Þ
effective, nonlinear, non-resonant susceptibility of
theinterface,emulatingthe substratesignal. Theangle
A.A. Mani et al./Applied Surface Science 237 (2004) 444–449445
brackets average the domain contributions, so that
only the xð2Þ
monolayer second-order susceptibility yield ppp SFG
signal.Inthe simulations, xð2Þ
the off-resonance SFG intensity in concordance with
the experimental lineshape of the modes, and the
vibration homogeneous broadening is set to 6 cm?1.
Because all the resonant and non-resonant contribu-
tions to the interface susceptibility interfere, SFG
resonances appear either as peaks or dips, or adopt
more complex shapes . In practice, we neglect the
influence of the adsorbed monolayer on the interface
Fresnel factors. This approach, truly valid in the low
coverage limit, must be used as a first approximation,
since the monolayer dielectric constant is unknown.
The TP and NAT vibrational first hyperpolarizabil-
ities were calculated ab initio. The NAT alkane chain
was truncated to make the calculations tractable. The
use of the shorter NO2–C6H4–NH–CH3 molecule
bears little influence on the SFG signature because
the highly symmetricand apolar chain presents aweak
SFG cross-section. The molecule geometries were
fully optimized in Gaussian  using the 6–311
þþG(d,p) (TP) or 6–311G** (NAT) atomic basis sets
within the hybrid B3LYP  DFT scheme (TP) or at
the restricted Hartree-Fock level (NAT). When the
residual forces were less than 10?5a.u., the harmonic
vibrational modes were determined analytically using
the coupled-perturbed Kohn-Sham approach .
Then, their dynamic dipoles @m=@Qv were found
analytically from the first-order nuclear coordinate
derivatives of the Kohn–Sham orbitals. The Raman
tensors @a=@Qvwere evaluated by a numerical differ-
entiation of the static polarizability tensor involving
four distorted molecule geometries per normal modes
xzx, and xð2Þ
zxxcomponents of the
nr isadjusted toreproduce
The SAM were produced by immersing polycrys-
talline Au in a 2mgmol/l solution of NAT in ethanol
. The substrates were mounted into a Kel-F spec-
tro-chemical cell sealed with a prismatic BaF2win-
dow. The NAT synthesis  and the procedure used
to control the film growth  are described else-
where. The NAT spectra were referenced to the spec-
trum of bare gold in pure ethanol.
The in situ SFG spectrum of NAT/Au evolves
gradually during film growth before stabilizing .
The final spectrum (Fig. 1) exhibits four resonances at
1327 cm?1(a), 1307 cm?1(b), 1281 cm?1(g), and
1187 cm?1(d). The first two are the strongest. The a
and d peaks have been assigned by ex situ infrared
absorption spectroscopy to the NO2symmetric stretch
of NATand to an in-plane deformation of the ring CH
bonds, respectively .
Our calculations predict 6 NAT vibrations between
1150 and 1340 cm?1, at 1329 cm?1(A mode),
1326 cm?1(B), 1312 cm?1(C), 1279 cm?1(D),
1191 cm?1(E), and 1173 cm?1(F). Due to the the-
oretical procedure limitations , the mode frequen-
spectra (Fig. 1). The calculations confirm that the a
resonance corresponds to the symmetric stretch of
NO2concomitant with an in-plane ring deformation
(A and Bmodes).The bvibrationmatches amode that
involves the NO2symmetric stretch as well (C mode).
The D, E, and F modes involve in-plane deformations
of the ring CH bonds and carbon skeleton. The g
A.A. Mani et al./Applied Surface Science 237 (2004) 444–449446
Fig. 1. In situ (bottom) and theoretical (top) SFG spectra of para-
nitro-anilino-dodecane-thiol SAM on Au, for ppp polarizations with
65?(IR) and 55?(visible) incidence angles. The substrate exposure
time to the NAT solution was 2 h. The continuous curve passing
throughthe experimentaldataresultsfrom a least squarefitbased on
a lorentzian model of the four SFG resonances . The NO2–
C6H4–NH–CH3moleculeon which the calculationswere performed
previous attribution .
All these vibrations present stronger components
for their infrared dipoles and Raman tensors along the
molecule vertical axis, joining the two N atoms (i.e.
bzzzdominates). Therefore, the molecule tilt u is
constrained to small angles to yield detectable SFG
intensity. Indeed, if the molecule cants towards the
surface, the vibrations are screened by the metal sur-
face, and the SFG signal behaves like ?cos6u due to
the Fresnel factors. The calculated intensity of the
strong a resonance drops rapidly beyond u ¼ 40?(to
15% of its maximum at 40?, and 0.8% at 60?), so that
the molecule tilt should likely be below ?40?to
enable the observation of the weaker modes. This
constraint agrees well with the value of 36?deduced
by SHG for the same surface in ethanol .
Unfortunately, all the NAT vibrations investigated
here present the same pattern for their infrared dipoles
and Raman tensors. As a result, the relative intensity
of the modes is largely insensitive to the molecule
orientation. In particular, the SFG spectra are virtually
independent of j when u<60?, as expected since
nearly all the molecular IR and Raman activity occurs
along the N–N axis. A typical spectrum is presented in
Fig. 1. The resolution of the precise molecule orienta-
tion requires thus to investigate spectral ranges where
the vibrational first hyperpolarizability has a richer
structure. This has been successfully achieved for
thiophenol adsorbed on Ag(1 1 1).
5. Thiophenol/Ag(1 1 1)
The thiophenol SAM were produced by immersing
Ag substrates for 24 h in solutions of 1.0 mM of TP in
ethanol. The predominantly Ag(1 1 1) surfaces were
formed by evaporating 200 nm of Ag on mica.
The ppp SFG spectrum of TP SAM recorded
between 500 and 1050 cm?1(Fig. 2) exhibits two
major resonances, around 690 and 1000 cm?1, in
agreement with previous experiments performed with
an OPO at 1000 cm?1 and a free electron laser .
According to our calculations, they must be attributed
to two in-plane deformations of the TP ring: the A
mode at 1014 cm?1and the E mode at 685 cm?1. The
major features of the dynamic dipoles and Raman
tensors of the most intense SFG-active vibrations
found in the 500–1050 cm?1spectral interval are
summarized in Table 1. They will allow us to establish
the TP orientation on the Ag(1 1 1) surface. Only the
ppp polarization signal was measured because the
strong screening of the parallel component of the
electric field on the Ag metal surface prevents the
use of the other polarization combinations to extract
the molecule orientation. (This is confirmed by our
theoretical SFG spectra of TP for the other polariza-
tions, which predict intensities below our detection
The two modes observed present a strong vibra-
A.A. Mani et al./Applied Surface Science 237 (2004) 444–449447
Fig. 2. Normalized experimental (lowest curve) and theoretical
SFG spectra of tiophenol adsorbed on Ag(1 1 1), for ppp polariza-
tions with 65?(IR) and 55?(visible) incidence angles. The middle
curve is the better fit (u ¼ 37?and j ¼ 0?) to experiment.
Dominant components of the vibration dynamic dipoles and Raman
tensors in TP, below 1050 cm?1(y: strong;§: weak)
Mode Frequency (cm?1)IR dipoleRaman tensor
along 0:5ex? 0:9ey
xx, ayy, ay
xx, ayy, ay
The ezdirection is the molecule vertical axis (the S–C bond), eyis
to the molecule axis. Theoretical frequencies were scaled by 0.97
share a similar shape of the Raman tensor around it.
On the other hand, their Raman activities differ very
significantly along that axis. Their relative SFG inten-
sity is thereforegovernedby the tilt of the molecule on
the surface: the intensity of the A mode is maximum
when the S–C bond is vertical, while the E mode
intensity should peak around 55?given that its Raman
and IR activities occur in perpendicular directions.
The comparison between the experimental and theo-
retical SFG spectra (Fig. 2) points to an angle u of
The D mode described in Table 1 corresponds to an
out-of-plane deformation of the ring CH bonds. It has
a much weaker Raman activity (mainly in the plane
containing the S–C axis and the ring normal). Contra-
rily to the A and E modes, its dipole moment is
oriented perpendicular to the TP ring. Consequently,
itsintensitydependson bothu andj,and,inparticular,
the flatter the moleculelies on the surface, the stronger
the D mode. Since this mode, and the more compli-
cated B and C modes, are not detected, the molecule
when the two angles are set to u ¼ 37?? 10?and
j ¼ 0?? 20?.
These findings agree well with previous angle-
resolved X-ray photoemission
Ag(1 1 1) that found u ¼ 24?? 5?. Theyconfirm
as well the smaller tilt of the molecule on Ag(1 1 1)
than on Au(1 1 1), where u?49?. Finally, the per-
pendicular orientation of the TP ring with respect to
the substrate surface (i.e. j ¼ 0?) matches that of the
herringbone packing, often adopted by adsorbed aro-
matic chromophores [20,21].
studies of TP/
The para-nitroanilino-dodecane-thiol and thiophe-
nol molecules form ordered films on Au and
Ag(1 1 1), respectively. The tilt angle of the NAT
molecule adsorbed on Au and studied in situ at the
end of the film growth is smaller than 60?and prob-
ably inferior or equivalent to ?40?. The comparison
of experimental and ab initio theoretical SFG spectra
of TP SAM indicates that the molecule S–C bond is
tilted ?37?from the surface normal,with its aromatic
ring oriented perpendicular to the surface. The accu-
racy on the TP orientation results from the vibrational
range inspected, between 500 and 1050 cm?1, where
the moleculevibrational first hyperpolarizability exhi-
bits pronounced spectral differences. Correspond-
ingly, the determination of the NAT molecule
orientation is impaired by the similar shape of the
hyperpolarizability tensor of the vibrations found
between 1150 and 1400 cm?1.
We thank Dr. G.-Q. Lu for assistance during the
SFG experiment and Dr. J.J. Wolff for preparing the
Associate (RA), Senior RA, Scientific Research
Worker, and Postdoctoral Researcher, respectively,
of the Belgian National Fund for Scientific Research
(FNRS). This work was supported by the Belgian
Fund for Joint Basic Research, by the Belgian Inter-
university Program on ‘‘Quantum size effects in
nanostructured materials’’ PAI/IUAP 5/1 initiated
by the Belgian Office for Scientific, Technical, and
Cultural Affairs (OSTC), and by the U.S. Department
of Energy, Division of Materials Sciences under
Award No. DEFG02-91ER45439, through the Freder-
ick Seitz Materials Research Laboratory at the Uni-
versity of Illinois at Urbana-Champaign.
 A.J. Heeger, Rev. Mod. Phys. 73 (2001) 681.
 A.G. MacDiarmid, Rev. Mod. Phys. 73 (2001) 701.
 P. Guyot-Sionnest, N. Cheuk-Yu (Eds.), Laser Techniques for
State-Selected and State-to-State Chemistry, Proc. SPIE, vol.
 R. Braun, B.D. Casson, C.D. Bain, E.W.M. van der Ham,
Q.H.F. Vrehen, E.R. Eliel, A.M. Briggs, P.B. Davies, J. Chem.
Phys. 110 (1999) 4634.
 A. Peremans, A. Tadjeddine, R. Prazeres, F. Glotin, D. Jar-
oszynski, J.-M. Ortega, Nucl. Instrum. Meth. A 341 (1994)
 A.A. Mani, L. Dreesen, P. Hollander, C. Humbert, Y. Caudano,
P.A. Thiry, A. Peremans, Appl. Phys. Lett. 79 (2001) 1945.
A.A. Gewirth, J.O. White, P.A. Thiry, A. Peremans, Opt. Lett.
29 (2004) 274.
 S.H. Lin, A.A. Villaeys, Phys. Rev. A 50 (1994) 5134.
 T.F. Heinz, H.-E. Ponath,G.I. Stegeman,in: NonlinearSurface
Electromagnetic Phenomena, Elsevier Science, Amsterdam,
A.A. Mani et al./Applied Surface Science 237 (2004) 444–449448
 L. Dreesen, C. Humbert, M. Celebi, J.-J. Lemaire, A.A. Mani, Download full-text
P.A. Thiry, A. Peremans, Appl. Phys. B 74 (2002) 621.
 M.J. Frisch, et al. GAUSSIAN 98, Revision A.1, Gaussian
Inc., Pittsburg, PA, 1998.
 A.D. Becke, J. Chem. Phys. 98 (1993) 5648.
 A. Komornicki, G.J. Fitzgerald, J. Chem. Phys. 98 (1993)
1398, and references therein.
 B. Champagne, Chem. Phys. Lett. 58 (1996) 261.
 C. Humbert, M. Buck, A. Calderone, J.-P. Vigneron, V. Meu-
nier, B. Champagne, W.-Q. Zheng, A. Tadjeddine, P.A. Thiry,
A. Peremans, Phys. Stat. Sol. (a) 175 (1999) 129.
 A. Wesch, O. Dannenberger, C. Wo ¨ll, J.J. Wolff, M. Buck,
Langmuir 12 (1996) 5330.
 A.P. Scott, L. Radom, J. Chem. Phys. 100 (1996) 16502.
 F. Eisert, O. Dannenberger, M. Buck, Phys. Rev. B 58 (1998)
 S. Frey, V. Stadleer, K. Heister, W. Eck, M. Zharnikov, M.
Grunze, B. Zeysing, A. Terfort, Langmuir 17 (2001) 2408.
 A.-A. Dhirani, R.W. Zehner, R.P. Hsung, P. Guyot-Sionnest,
L.R. Sita, J. Am. Chem. Soc. 118 (1996) 3319.
 S.-C. Chang, I. Chao, Y.-T. Tao, J. Am. Chem. Soc. 116 (1994)
A.A. Mani et al./Applied Surface Science 237 (2004) 444–449449