Periodic trends in electrode-chemisorbate bonding: Benzonitrile on platinum-group and other noble metals as probed by surface-enhanced Raman spectroscopy combined with Density Functional Theory

Article abstract of DOI:10.1021/ja010049k

Detailed intramolecular vibrational spectra obtained by means of surface-enhanced Raman scattering (SERS) for benzonitrile adsorbed on seven electrode surfaces-four Pt-group metals (platinum, palladium, rhodium, and iridium) and the Group IB metals (coppe

Full text of DOI:10.1021/ja010049k

J. Am. Chem. Soc. 2001, 123, 12817-12825
12817
Periodic Trends in Electrode-Chemisorbate Bonding: Benzonitrile
on Platinum-Group and Other Noble Metals As Probed by
Surface-Enhanced Raman Spectroscopy Combined with Density
Functional Theory
Melissa F. Mrozek, Sally A. Wasileski, and Michael J. Weaver*
Contribution from the Department of Chemistry, Purdue UniVersity, West Lafayette, Indiana 47907-1393
ReceiVed January 4, 2001
Abstract: Detailed intramolecular vibrational spectra obtained by means of surface-enhanced Raman scattering
(SERS) for benzonitrile adsorbed on seven electrode surfacessfour Pt-group metals (platinum, palladium,
rhodium, and iridium) and the Group IB metals (copper, silver, and gold)sare reported with the aim of exploring
the metal-dependent nature of surface-chemisorbate interactions. The Pt-group surfaces were prepared as
ultrathin electrodeposited films on gold, enabling the SERS activity inherent to the substrate to be imparted to
the overlayer material. Benzonitrile was selected as a “model” organic adsorbate since it displays a rich array
of coupled aromatic ring as well as substituent modes which collectively can provide insight into the various
molecular perturbations induced by surface coordination via the nitrile substituent. The experimental spectra
are compared with ab initio calculations of vibrational frequencies, bond geometries, and charge distributions
obtained by means of Density Functional Theory (DFT), which yields valuable insight into the underlying
structural reasons for the sensitivity of the experimental coordination-induced frequency shifts to the nature of
the intramolecular mode and the metal surface. The DFT results also form an invaluable aid in making SER
spectral assignments, along with providing detailed information on the coupled atomic displacements involved
in each vibrational mode. Benzonitrile surface coordination was modeled in the DFT calculations by binding
the nitrile group to metal atoms and small metal clusters. While the majority of the aromatic-ring SER frequencies
are altered only slightly (j5 cm-1) upon surface coordination, several modes (especially ν1, ν6a) are blue-
shifted substantially (by up to 50 cm-1). These shifts were identified by DFT as arising from mode coupling
to the nitrile substituent, especially involving the C-CN bond that is compressed upon nitrile coordination,
associated with metal-adsorbate back-donation. The small (<5 cm-1) red-shifts seen for ring vibrations not
involving coupled substituent motion apparently arise from increased antibonding aromatic electron density.
The metal-dependent frequency shifts seen for these coupled aromatic vibrations as well as for the more localized
C-N nitrile stretching mode are consistent with increased back-donation anticipated in the sequence d10 < d9
< d8 within a given Periodic row. Overall, the findings provide a benchmark illustration of the virtues of DFT
in interpreting complex vibrational spectra for larger polyatomic adsorbates.
Introduction
(SERS), offers typically higher sensitivity, more liberal selection
rules, and a wider effective wavenumber range. As a conse-
quence, SERS often yields much richer vibrational spectra than
IRAS, including the ability to detect surface-adsorbate and
other low-frequency vibrations, and the common appearance
of most (if not all) normal modes of polyatomic molecules.
Indeed, these attributes of SERS are in some respects compa-
rable to those of electron energy-loss spectroscopy (EELS),
although the latter suffers from poorer wavenumber resolution
and is only applicable to UHV-based interfaces. However, the
usual need to employ coinage-metal substrates would appear
to limit severely the types of interfaces amenable to character-
ization by SERS.
Understanding how the nature of chemisorbate binding
depends on the chemical nature of the metal surface constitutes
an issue of major fundamental significance in electrochemistry
as well as in ultrahigh vacuum (UHV)-based surface science.
For the latter type of interface, experimental information
regarding metal-chemisorbate interactions is attainable in
several ways, including temperature-dependent desorption along
with electronic and vibrational spectroscopies. For in situ
electrochemical interfaces, however, the breadth of available
molecular-level probes is considerably narrower. Although
infrared reflection-absorption spectroscopy (IRAS) has found
extensive application at electrochemical (EC) as well as metal-
UHV interfaces, the limitations set by sensitivity, selection rules,
and solution-phase interferences place restrictions on the type
of vibrational modes (and hence the range of adsorbate
molecules) suitable for characterization by EC-IRAS. The other
type of vibrational technique commonly applied to electro-
chemical interfaces, surface-enhanced Raman spectroscopy
Mindful of the unique virtues of SERS for the detailed
vibrational characterization of metal-ambient interfaces, we
have long been interested in expanding its applicability to
surface materials beyond the coinage metals.1-4 Our primary
(1) For a recent overview, see: Weaver, M. J.; Zou, S.; Chan, H. Y. H.
Anal. Chem. 2000, 72, 38A.
(2) (a) Leung, L.-W. H.; Weaver, M. J. J. Am. Chem. Soc. 1987, 109,
5113. (b) Leung, L.-W. H.; Weaver, M. J. Langmuir 1988, 4, 1076.
* Address correspondence to this author. E-mail: mweaver@purdue.edu.
10.1021/ja010049k CCC: $20.00 © 2001 American Chemical Society
Published on Web 12/01/2001

12818 J. Am. Chem. Soc., Vol. 123, No. 51, 2001
Mrozek et al.
tactic involves electrodepositing the material of interest as an
ultrathin (nanoscale) film onto gold.1 This exploits the chemical
(and electrochemical) inertness as well as excellent SERS
properties of gold, along with the demonstrated ability to impart
substantial Raman enhancement to species bound to the outer
surface of (and also within) the overlayer film. The strategy
has been utilized in particular to explore chemisorption on
platinum-group metal surfaces in elevated-temperature gaseous
as well as in electrochemical environments, prompted by their
catalytic importance, although a variety of other interfaces have
also been scrutinized in this manner, including oxides and
semiconductors.3 The range of adsorbates examined at Pt-group
electrodes with our “overlayer-SERS” strategy was limited
originally by the presence of residual exposed gold sites, which
can yield spectral and electrochemical interferences. More
recently, however, we have devised modified electrodeposition
procedures that yield ultrathin (3-5 monolayer) Pt-group metal
overlayers displaying optimal SERS properties that are es-
sentially pinhole-free, and thereby devoid of substrate interfer-
ences.6 While thicker films yield progressively weaker SERS
signals, a significant and eventually dominant contribution to
the Raman enhancement has been shown to emanate from the
transition-metal overlayer itself.7 These developments therefore
enable us to acquire vibrational spectra for a much wider range
of adsorbates at transition-metal electrochemical interfaces,
including species that bind also to gold electrodes. Indeed, we
have recently utilized this tactic to obtain potential-dependent
vibrational spectra at Pt-group electrodes for a variety of
chemisorbates, including halides,8a pseudohalides,8b sulfur,8a
oxides,9 alkenes,10 and aromatic molecules.11
As a consequence, SERS can now be harnessed to explore
in broad-based fashion the sensitivity of chemisorbate vibrational
properties to the electrode material. We recently examined
periodic trends in monatomic chemisorbate bonding on four
Group VIII metals (Pt, Pd, Ir, and Rh) along with the three
Group IB metals (Cu, Ag, Au) by monitoring the SERS
surface-adsorbate stretching mode as a function of electrode
potential.8 The rich vibrational spectra attainable by SERS for
aromatic chemisorbates furnishes a more complex, yet chemi-
cally important, class of systems with which to explore such
periodic trends. We have chosen benzonitrile (C6H5CN) for
initial detailed examination along these lines. Earlier SERS
studies on gold12 and palladium11 have shown that benzonitrile
binds to these metals via the nitrile substituent. While some
Raman bands associated with the pendant aromatic ring are
virtually unaltered in frequency (and band shape) from those
for uncoordinated benzonitrile, other ring modes (along with
the C-N vibration for the surface-attached nitrile group) are
shifted to an extent that is sensitive to the electrode material.
The present report provides a detailed SERS-based analysis
of the metal-dependent vibrational properties of benzonitrile
chemisorbed on seven electrode surfaces, specifically for four
Pt-group metals (Pt, Pd, Ir, and Rh) along with the three coinage
metals (Cu, Ag, and Au). We have also undertaken ab initio
calculations of the benzonitrile vibrational frequencies by means
of Density Functional Theory (DFT), with the objective of
elucidating the underlying structural reasons for the sensitivity
of the chemisorption-induced frequency shifts to the nature of
the intramolecular mode and the metal surface. The findings
provide a benchmark illustration of the value of DFT for aiding
the interpretation of the rich vibrational spectra attainable for
larger polyatomic adsorbates by using SERS, as well as
furnishing detailed insight into the relation between the vibra-
tional properties and the nature of electrode-chemisorbate
bonding.
Experimental and Computational Procedures
The experimental arrangement used for SERS is detailed in ref 15.
The Raman excitation was from a Spectra Physics Stabilite model 2017
Kr+ laser operated at 647.1 nm, with ca. 30 mW incident power focused
to a 1 mm spot on the electrode surface. Scattered light was collected
with a SPEX Triplemate spectrometer equipped with a Photometrics
PM 512 CCD detector. The gold, silver, and copper electrodes were
of rotating-disk construction, consisting of 2-4 mm diameter disks
sheathed in Teflon, polished with 1.0 and 0.3 µm alumina and rinsed
before use. Electrochemical roughening of the gold surface to yield
optimal SERS activity consisted of successive oxidation-reduction
cycles in 0.1 M KCl as outlined in ref 16. Similar electrochemical
roughening procedures were performed for silver and copper electrodes
as described in refs 17 and 18, respectively. Transition-metal SERS-
active surfaces were prepared by constant-current electrodeposition onto
a roughened gold electrode following the procedures outlined in ref 6.
Typically, 3-5 monolayers were deposited, using either an acidic
medium, 0.1 M HClO4 (for Pd and Rh), or a phosphate buffer solution,
0.7 M Na2HPO4 (for Pt and Ir).6 Electrolytes were prepared with
ultrapure water from a Millipore MilliQ system. All measurements were
made at room temperature (23 ( 1 °C), and all electrode potentials
are reported versus a saturated calomel electrode (SCE).
(3) (a) Desilvestro, J.; Corrigan, D. A.; Weaver, M. J. J. Phys. Chem.
1986, 90, 6408. (b) Zou, S.; Weaver, M. J. J. Phys. Chem. B 1999, 103,
2323.
(4) (a) Leung, L.-W. H.; Weaver, M. J. J. Electroanal. Chem. 1987, 217,
367. (b) Leung, L.-W. H.; Gosztola, D.; Weaver, M. J. Langmuir 1987, 3,
45.
(5) For example: (a) Williams, C. T.; Takoudis, C. G.; Weaver, M. J. J.
Phys. Chem. B 1998, 102, 406. (b) Tolia, A.; Williams, C. T.; Takoudis,
C. G.; Weaver, M. J. J. Phys. Chem. 1995, 99, 4599.
(6) (a) Zou, S.; Weaver, M. J. Anal. Chem. 1998, 70, 2387. (b) Zou, S.;
Go´mez, R.; Weaver, M. J. Langmuir 1997, 13, 6713. (c) Mrozek, M. F.;
Xie, Y.; Weaver, M. J. Anal. Chem. 2001, 74, 5953.
Bis(benzonitrile)dichloroplatinum was synthesized as described in
ref 19. PtCl2 (Aldrich) was dissolved in a minimum volume of
benzonitrile at 100 °C and stirred for 1 day. The yellow precipitate of
cis-PtCl2(C6H5CN)2 was obtained after the solution was cooled and
filtered. More product could be obtained by diluting the filtrate with
light petroleum (bp 40-60 °C). The product was recrystallized in
benzene, washed with water and ethanol, and dried in vacuo.
The DFT calculations reported here model chemisorbed benzonitrile
by binding the lead-in nitrogen to three types of atop metal sites. The
chief approach used involves only a single metal atom. While clearly
not a quantitative model of a metal surface (!), as noted below this
(7) Zou, S.; Weaver, M. J.; Li, X. Q.; Ren, B.; Tian, Z. Q. J. Phys.
Chem. B 1999, 103, 4218.
(8) (a) Mrozek, M. F.; Weaver, M. J. J. Am. Chem. Soc. 2000, 122, 150.
(b) Luo, H.; Weaver, M. J. Langmuir 1999, 15, 8743.
(9) (a) Chan, H. Y. H.; Zou, S.; Weaver, M. J. J. Phys. Chem. B 1999,
103, 11141. (b) Zou, S.; Chan, H. Y. H.; Williams, C. T.; Weaver, M. J.
Langmuir 2000, 16, 754.
(10) Mrozek, M. F.; Weaver, M. J. J. Phys. Chem. B. 2001, 105, 8931.
(11) (a) Zou, S.; Williams, C. T.; Chen, E. K.-Y.; Weaver, M. J. J. Am.
Chem. Soc. 1998, 120, 3811. (b) Zou, S.; Williams, C. T.; Chen, E. K.-Y.;
Weaver, M. J. J. Phys. Chem. B 1998, 102, 9039 [also see erratum: Zou,
S.; Williams, C. T.; Chen, E. K.-Y.; Weaver, M. J. J. Phys. Chem. B 1998,
102, 9743].
(13) Gao, X.; Zhang, Y.; Weaver, M. J. J. Phys. Chem. 1992, 96, 4156.
(14) Gao, P.; Gosztola, D.; Leung, L.-W. H.; Weaver, M. J. J.
Electroanal. Chem. 1987, 233, 211.
(15) (a) Wilke, T.; Gao, X.; Takoudis, C. G.; Weaver, M. J. J. Catal.
1991, 130, 62. (b) Gao, X.; Zhang, Y.; Weaver, M. J. Langmuir 1992, 8,
688.
(16) Gao, P.; Gosztola, D.; Leung, L. H.; Weaver, M. J. J. Electroanal.
Chem. 1987, 233, 211.
(17) Leung, L. H.; Gosztola, D.; Weaver, M. J. Langmuir 1987, 3, 45.
(18) Chan, H. Y. H.; Takoudis, C. G.; Weaver, M. J. J. Phys. Chem. B
1999, 103, 357.
(12) (a) Gao, P.; Weaver, M. J. J. Phys. Chem. 1985, 89, 5040. (b) Gao,
X.; Davies, J. P.; Weaver, M. J. J. Phys. Chem. 1990, 94, 6858.
(19) Hartley, F. R. The Chemistry of Palladium and Platinum; Applied
Science: London, 1973; p 462.

Periodic Trends in Electrode-Chemisorbate Bonding
J. Am. Chem. Soc., Vol. 123, No. 51, 2001 12819
arrangement mimics adequately the coordinative effects upon the
aromatic intramolecular bonding of interest here. Nevertheless, DFT
calculations were also undertaken by using two simple metal-cluster
models. The first involves a three-atom triangular cluster in which the
benzonitrile aromatic ring is oriented in the same plane, and the second
utilizes a five-atom single-layer (100) cluster in which the nitrile is
bound to the center atom. The metal-metal distances were fixed at
their bulk-phase experimental values. While satisfactory convergence
was not attained with calculations for the Pt-group coordinated clusters,
reliable bond-length and vibrational-frequency data were extracted from
the DFT calculations for the Au3 cluster. Representative DFT results
are therefore included below for this system. (Comparable benzonitrile
bond-length data were obtained for the Au5 and Au3 clusters, although
satisfactory vibrational frequency results were not achieved for the
former system.)
While clearly less appropriate than utilizing larger metal clusters,
the present approach has the advantage of computational efficiency in
addition to maintaining C2V symmetry, so that the individual calculated
harmonic vibrational modes of the coordinated molecule can readily
be compared to those of the uncoordinated molecule. All calculations
employed the Amsterdam Density Functional (ADF) package,20 using
the BLYP functional21 for the generalized gradient approximation and
the Vosko-Wilk-Nusair form of the local density approximation,22
with Slater-type atomic orbital basis sets. The inner-core electrons of
all atoms, except H, were kept frozen to the following orbitals: C 1s,
N 1s, Au 4f, Pt 4f, and Ir 4f. All basis sets were of double-ú quality
and C, N, and H were augmented by polarization functions. The
benzonitrile bond lengths and bond angles extracted from ref 23 were
used as the initial input for the geometry optimization, which employed
convergence criteria of 10-4 and 10-3 hartrees for the energy and
gradient, respectively, in determining the calculated equilibrium
geometry. Harmonic vibrational frequencies were calculated both for
this equilibrium geometry and with the metal-nitrogen bond fixed
instead at 2.00 Å (vide infra) using 2-point numerical differentiation,
in which negative and positive bond-length displacements are used to
determine the energy gradients and compute the force constants.
Results and Discussion
Vibrational Frequencies and Mode Assignments. Repre-
sentative surface-enhanced Raman (SER) spectra obtained for
benzonitrile on the three Group IB metals (Cu, Ag, Au) at 0 V
vs SCE, adsorbed from a 20 mM solution in 0.1 M HClO4, are
shown in Figure 1A. Corresponding spectra obtained on the four
Group VIII metals (Rh, Ir, Pd, Pt) are illustrated in Figure 1B.
A detailed summary of the resulting SER band frequencies for
benzonitrile adsorbed on these seven metal surfaces is given in
Table 1, along with the corresponding experimental values,
ν(exp), for uncoordinated (liquid) benzonitrile. The latter were
also chiefly as measured in this work, supplemented by literature
values (given in parentheses).24,25 The band assignments (num-
bered with the Wilson notation) are also identified in Table 1
by means of an approximate mode description. While the
qualitative physical nature of most of these normal modes has
been described in analyses by Varsanyi,24 the present DFT
calculations also yield a complete picture of the vibrational
(20) (a) Amsterdam Density Functional Package, ADF 2000.03, Depart-
ment of Theoretical Chemistry, Vrije Universiteit, Amsterdam, The
Netherlands, 2000. (b) Baerends, E. J.; Ellis, D. E.; Ros, R. Chem. Phys.
1973, 2, 41. (c) teVelde, G.; Baerends, E. J. J. Comput. Phys. 1992, 99, 84.
(21) Becke, A. D. Phys. ReV. A 1988, 38, 3098. (b) Lee, C.; Yang, W.;
Parr, P. G. Phys. ReV. B 1988, 37, 785.
(22) Vosko, S. H.; Wilk, L.; Nusair, M. Can. J. Phys. 1980, 58, 1200.
(23) Bak, B.; Christensen, D.; Dixon, W. B.; Hansen-Nygaard, L.;
Rostrup-Anderson, J. J. Chem. Phys. 1962, 37, 2027.
(24) (a) Varsa´nyi, G. Vibrational Spectra of Benzene DeriVatiVes;
Academic Press: New York, 1969. (b) Varsa´nyi, G. Assignments for
Vibrational Spectra for 700 Benzene DeriVatiVes; Wiley: New York, 1974.
(25) (a) Green, J. H. S. Spectrochim. Acta 1961, 17, 607. (b) Jakobsen,
R. J. Spectrochim. Acta 1964, 21, 127. (c) Green, J. H. S.; Harrison, D. J.
Spectrochim. Acta 1976, 32A, 1279.
Figure 1. Surface-enhanced Raman spectra obtained for benzonitrile
on (A) copper, silver, and gold and (B) rhodium, iridium, palladium,
and platinum at 0 V vs SCE from 20 mM solution in 0.1 M HClO4.

12820 J. Am. Chem. Soc., Vol. 123, No. 51, 2001
Mrozek et al.
Table 1. Raman Frequencies (cm-1) for Benzonitrile Adsorbed on Various Electrode Surfaces
uncoordinated
SERSf
Pd
descriptiona
modeb
symc
a1
ν(DFT)d
ν(exp)e
Cu
Ag
Au
Pt
Rh
Ir
C-C-C in-plane bend
ring breathing
C-C-C trigonal breathing
C-H in-plane bend
C-H in-plane bend
C-X stretch
C-C in-plane stretch
C-C in-plane Stretch
C-C-C out-of-plane bend
C-H out-of-plane bend
C-C-C out-of-plane bend
C-H out-of-plane bend
C-H out-of-plane bend
C-C-C in-plane bend
C-H in-plane bend
C-H in-plane bend
C-C in-plane stretch
C-C in-plane stretch
CN stretch
ν6a
ν1
463
763
462
768
504
786
480
775
485
772
493
781
514
778
501
780
997
512
788
995
ν12
ν18a
ν9a
1002
1030
1181
1200
1491
1600
394
1002
1027
1180
1194
1490
1599
(400)
(848)
551
754
(926)
626
1162
(1290)
(1336)
1449
2232
999
999
998
999
1000
1027
1178
1196
1490
1593
1023
1178
1199
1486
1593
1025
1177
1197
1488
1595
385
1026
1174
1194
1484
1592
380
1024
1175
1194
1484
1593
395
1024
1175
1198
1484
1591
398
1022
1173
1191
1486
1591
398
ν7a
ν19a
ν8a
ν16a
ν10a
ν16b
ν11
ν17b
ν6b
ν18b
ν3
a2
b1
833
850
850
850
542
765
927
626
850
835
546
555
762
932
622
550
758
930
626
545
542
536
544
757
762
765
765
768
915
921
934
933
922
b2
633
624
628
626
626
1169
1310
1337
1445
2247
1290
1392
1444
2246
[10]
1298
1375
1440
2240
[17]
1280
1375
1440
2233
[14]
1270
1375
1440
2228
[18]
ν14
ν19b
νCN
1390
1375
1375
1445
2226
[16]
a1
2242
[13]
2229
[9]
a Approximate mode description, extracted chiefly from discussions in ref 24a, along with DFT analysis (see text). b Wilson notation for normal
vibration, from ref 24. c Symmetry group of vibration. d Vibrational frequencies calculated by DFT, scaled by 0.975 (so to align with experimental
values, see text). e Raman frequencies for liquid benzonitrile, from present measurements, except those in parentheses, which were extracted from
refs 24 and 25. f SER frequencies for benzonitrile adsorbed at 0 V vs SCE on the metal surface indicated. The values given in brackets underneath
the νCN frequency for each metal are the “Stark-tuning” slopes, dνCN/dE (cm-1 V-1), evaluated between -0.2 and 0.4 V on the Pt metals, -0.4 and
0 V on Cu and Ag, and -0.4 and 0.4 V on Au.
x-y-z motion of each atom away from equilibrium. We
consider below more detailed descriptions for selected modes.
The utility of DFT in this regard is highlighted by the ability
of the BLYP functional to predict accurately (chiefly within
5-10 cm-1) the benzonitrile mode frequencies. The computa-
tional frequencies, ν(DFT), also listed in Table 1, were adjusted
slightly (with a 0.975 scaling factor) to optimize agreement with
the experimental values for liquid benzonitrile, ν(exp), given
alongside. Matching up the ν(exp) and ν(DFT) values, along
with the vibrational mode intensities, enables complete mode
assignments to be made with confidence.
We have previously analyzed in detail SER spectra for
benzonitrile on gold,12 with a similar study described more
recently on palladium.11b,26 The combined frequency-band
shape-intensity analysis, the last by utilizing surface selection
rules,12b indicates a benzonitrile orientation normal (or tilted)
to the metal surface, being attached via the nitrile group. The
absence of substantial aromatic-surface interactions on gold
was also evident from the narrow (fwhm ∼ 10 cm-1) band-
widths for most aromatic ring modes (similar to the bulk
liquid).12 The nitrile stretching (νCN) vibration, however, is
substantially broadened and shifted in frequency upon adsorp-
tion, indicative of direct metal-nitrile binding (Figure 1). These
spectral properties, as discussed previously for adsorption on
gold12 and palladium,11b,26 sharply contrast those observed for
another monosubstituted benzene, toluene (C6H5CH3), on the
same surfaces: the latter exhibits ring vibrations that are
broadened as well as shifted significantly (by ca. 10-50 cm-1).
Aside from constituting direct evidence for the presence of
strong metal-ring interactions for toluene, consistent with the
anticipated flat surface orientation, the contrasting spectral
features for benzonitrile indicate further the absence of this
surface bonding mode for the latter adsorbate.
trodes at more negative potentials.27 A similar orientation has
been deduced from electron energy loss spectroscopy (EELS)
for benzonitrile on Cu(111)28 and on Au(100)29 in UHV. (Note,
however, that the “flat” orientation is apparently observed even
for cyanogen at clean metal-UHV interfaces, rather than the
“end-on” binding observed for -CN at metal electrodes,
presumably reflecting the role of solvation and other factors.30)
Analyses of SERS data for benzonitrile on silver sols31 and
electrodes32 suggest either “end-on” or “face-on” orientations,
with the nitrile group nonetheless providing the primary surface
anchor. The original SERS report of benzonitrile adsorption on
gold mentioned the appearance of low-frequency “shoulders”
on the ν12 and ν18a features at negative electrode potentials,12
suggestive of more direct surface-ring interactions (i.e., a flat
orientation) under these conditions.
However, no clear-cut evidence for such potential-induced
changes in the mode of benzonitrile surface binding, or the
nature of the surface-adsorbate interactions, was obtained in
the present study. While some details differ, the sharp ring-
mode bands along with the broadened νCN feature attest to the
likely similarity of nitrile surface attachment on each metal.
While the spectra shown in Figures 1A and 1B refer to 0 V vs
SCE, data were also obtained between -0.3 and 0.4 V in 0.1
M HClO4, and down to -0.8 V in 0.1 M NaClO4 (pH 11), to
examine any sensitivity to the electrode potential. Most of the
vibrational band frequencies listed in Table 1 do not shift
perceptibly (e2 cm-1) over the potential range -0.3 to 0.4 V,
commonly assessable with 0.1 M HClO4 electrolyte. There are
two exceptions. As reported previously on gold,12 the nitrile
(νCN) vibration blue-shifts toward higher potentials: the
(27) Richer, J. F.; Chen, A.; Lipkowski, J. Electrochim. Acta 1998, 44,
1037.
(28) Kordesch, M. E.; Feng, W.; Stenzel, W.; Weaver, M. J.; Conroad,
H. J. Electron Spectrosc. Related Phenom. 1987, 44, 149.
(29) Solomun, T.; Christmann, D.; Baumgartel, H. J. Phys. Chem. 1989,
93, 7199.
(30) Kordesch, M. E.; Stenzel, W.; Conrad, H.; Weaver, M. J. J. Am.
Chem. Soc. 1987, 109, 1878.
(31) Joo, T. H.; Kim, K.; Kim, H.; Kim, M. S. Chem. Phys. Lett. 1985,
117, 518.
(32) Holze, R. Electroanalysis 1993, 5, 497.
Some thermodynamic and infrared spectral evidence, how-
ever, suggests that benzonitrile binds “flat” on Au(111) elec-
(26) The SER spectrum reported originally in ref 11b for benzonitrile
adsorbed on a palladium film displayed Raman bands that were broadened
artifactually by the use of a wide spectrometer band-pass. A corrected
spectrum, however, is reported in the erratum to that paper.11b

Periodic Trends in Electrode-Chemisorbate Bonding
Table 2. Selected Vibrational Frequency Shifts (cm-1) for Benzonitrile Adsorbed on Various Electrode Surfaces
frequency shiftsb ∆ν(exp), cm-1
J. Am. Chem. Soc., Vol. 123, No. 51, 2001 12821
description
mode
sym
a1
liqa
Cu
Ag
Au
Pd
Pt
Rh
Ir
C-C-C in-plane bend
ring breathing
C-C-C trigonal breathing
C-H in-plane bend
C-H in-plane bend
C-X stretch
C-C in-plane stretch
C-C in-plane stretch
C-H out-of-plane bend
C-C-C in-plane bend
CN stretch
ν6a
ν1
462
768
+42
+18
-3
+18
+7
-3
-2
-3
+3
-2
-4
+4
0
+23
+4
-4
-1
-4
0
-6
-7
+8
-2
+8
+31
+13
-3
-3
-5
0
-6
-6
+11
+2
-3
+52
+10
-2
0
-2
+2
0
-6
+11
0
+39
+12
-5
-3
-5
+4
-6
-8
+11
0
+50
+20
-7
-5
-7
-3
-4
-8
+14
0
ν12
ν18a
ν9a
ν7a
ν19a
ν8a
ν11
ν6b
νCN
1002
1027
1180
1194
1490
1599
754
-3
-2
+5
-4
-6
b1
b2
a1
+8
626
2232
-4
+14
+10
+1
-6
-4
a Raman frequencies measured for liquid benzonitrile (see Table 1). b Frequency shifts observed upon benzonitrile adsorption on the metal electrode
at 0 V vs SCE (data from Table 1).
Table 3. Comparison of Vibrational Frequencies (cm-1) of Bis(benzonitrile)dichloroplatinum and Benzonitrile Adsorbed on Platinum
b
description
mode
sym
a1
liqa
PtCl2(BN)2
∆νc(exp)c
Pt SERSd
∆νa(exp)e
C-C-C in-plane bend
ring breathing
C-C-C trigonal breathing
C-H in-plane bend
C-H in-plane bend
C-X stretch
C-C in-plane stretch
C-C in-plane stretch
C-H out-of-plane bend
C-C-C in-plane bend
C-H in-plane bend
CN stretch
ν6a
ν1
462
768
555
793
999
1027
1179
1204
1490
1594
763
+93
+25
-3
0
-1
+10
0
-5
+9
-2
+3
+62
514
778
+52
+10
-2
0
-2
+2
0
-6
+11
0
ν12
ν18a
ν9a
ν7a
ν19a
ν8a
ν11
ν6b
ν18b
νCN
1002
1027
1180
1194
1490
1599
754
626
1162
2232
1000
1027
1178
1196
1492
1593
765
b1
b2
624
1165
2294
626
a1
2233
+1
a Raman frequencies for liquid benzonitrile (see Table 1). b Raman frequencies for solid bis(benzonitrile)dichloroplatinum. c Frequency shift
upon Pt-benzonitrile complexation. d SER frequencies for benzonitrile adsorbed on Pt at 0 V vs SCE (from Table 2). e Frequency shift upon
benzonitrile adsorption on Pt.
dνCN/dE values, between 10 and 18 cm-1 V-1, are given in
brackets underneath the corresponding νCN frequencies in Table
1. The ν6a ring mode also blue-shifts (albeit only slightly) toward
higher potentials, with dν6a/dE ≈ 3-7 cm-1 V-1. An additional
νCN band appears at lower frequencies, for example at 2100-
2120 cm-1 on gold and 2035-2055 cm-1 on platinum, in the
alkaline electrolyte at more negative potentials, below -0.2 V.
While the band position is consistent with the occurrence of a
π-bonded nitrile group,33 and hence hinting at a “flat” adsorbate
orientation, the absence of this feature when using acidic and
neutral electrolytes suggests that it arises instead from cyanide
or related species formed by adsorbate hydrolysis.
To facilitate examination of the metal-dependent behavior,
Table 2 summarizes the frequency shifts, ∆ν(exp), observed
for selected modes upon binding benzonitrile to the seven metal
surfaces. These eleven modes were chosen in view of their ease
of experimental identification (from band intensities, band-
widths, etc). In particular, they include modes that exhibit
significant ∆ν(exp) values along with sensitivity to the metal
surface. Prominent among these are the ν6a band, which is seen
to blue-shift substantially, by ca. 20-50 cm-1, upon chemi-
sorption, along with the ν1 feature, which also blue-shifts
noticeably, by 10-20 cm-1 depending on the metal. It is
important to note that, although these modes were identified in
our earlier studies,11b,12 more accurate metal-dependent ∆ν(exp)
values have been obtained in the present work. (Aside from
the present use of higher spectrometer resolution, the necessary
matching of band assignments for chemisorbed and uncoordi-
nated benzonitrile is aided considerably by the present DFT
results, vide infra.) Interestingly, these ring-mode frequency
shifts are comparable to, or even larger than, those for the νCN
mode, which exhibits blue-shifts on the coinage metals, and
largely yields red-shifts on the Pt-group surfaces. The remaining
a1 symmetry modes display smaller (<10 cm-1) and metal-
insensitive ∆ν(exp) values. Of the clearly identifiable b sym-
metry modes, the ν11 vibration is significantly red-shifted upon
adsorption, especially on the Pt-group metals, while ν6b is
virtually unaltered (Table 2).
A basic question is whether these mode-sensitive frequency
shifts reflect the occurrence of “direct” surface-ring interactions,
or rather can be accounted for in terms of an inductive effect
via the nitrile substituent. It is therefore of interest to elucidate
whether the frequency shifts are inherent to surface coordination,
or if they occur also for metal-benzonitrile molecular com-
plexes. Especially given that the majority of the present DFT
calculations similarly refer to benzonitrile-metal atom binding,
we therefore have examined Raman vibrational shifts for
benzonitrile coordinated in a mononuclear Pt complex, bis-
(benzonitrile)dichloroplatinum [PtCl2(BN)2]. The chief Raman
frequencies observed for the solid-state sample are summarized
in Table 3. Listed alongside are the values for benzonitrile
adsorbed on platinum at 0 V, together with the resulting
frequency shifts in the complexed and adsorbed states, ∆νc-
(exp) and ∆νa(exp), respectively. Significantly, there are close
parallels in the ∆νc(exp) and ∆νa(exp) values. Thus both
coordination environments yield large blue-shifts in the ν6a
mode, and to a lesser extent for the ν1 and ν11 modes, with
comparable red-shifts in the ν8a vibration. The only clear
behavioral difference is seen for the νCN vibration, where the
complex, but not the adsorbate, exhibits a substantial blue-shift.
This difference is most simply attributable to the electropositive
(33) Nakamoto, K. Infrared and Raman Spectra of Inorganic and
Coordination Compounds, 5th ed.; Wiley: New York, 1997; Part B, pp
113-4.

12822 J. Am. Chem. Soc., Vol. 123, No. 51, 2001
Mrozek et al.
Table 4. Selected Vibrational Frequency Shifts (cm-1) for Benzonitrile Bound to Metal Atoms, as Calculated by DFT
frequency shiftsb, ∆ν(DFT), cm-1
Ir
Pt
Au
Au3
description
mode
ν(DFT)a cm-1
1.80 Å
2.00 Å
1.79 Å
2.00 Å
2.28 Å
2.00 Å
2.00 Å
C-C-C in-plane bend
ring breathing
C-C-C trigonal breathing
C-H in-plane bend
C-H in-plane bend
C-X stretch
C-C in-plane stretch
C-C in-plane stretch
C-H out-of-plane bend
C-C-C in-plane bend
C-H in-plane bend
CN stretch
ν6a
ν1
463
763
+141
+66
-5
+35
+12
-6
+136
+63
-3
-2
0
+30
-5
-6
-7
-4
-2
+1
+27
+12
-3
+9
+2
-1
-1
+1
+2
-2
-2
-5
-3
+3
-3
+53
+15
-1
0
+1
+3
-3
-5
-5
-3
+3
-2
+46
+17
-3
ν12
ν18a
ν9a
ν7a
ν19a
ν8a
ν11
ν6b
ν18b
νCN
1002
1030
1181
1200
1491
1600
757
633
1169
2247
-2
-2
-3
-2
0
-4
-1
+1
+32
+13
-6
+12
-5
+15
-2
-5
-8
-7
-6
-2
-8
-7
-6
-5
-3
-2
-4
-5
-3
-1
-1
+4
-28
-71
-41
+16
a Mode frequency for uncoordinated benzonitrile as calculated by DFT, scaled slightly (by 0.975) to match experimental values (from Table 1).
b Frequency shifts observed upon coordination of the nitrile nitrogen to the metal atom or the cluster indicated. The left-hand column for each metal
lists ∆ν(DFT) values obtained for equilibrium metal-N bond distance computed by DFT, as noted in the column header. The corresponding
right-hand column for each metal gives ∆ν(DFT) values calculated for a metal-N bond distance constrained at 2.00 Å (see text).
nature of Pt(II), which should enhance the degree of nitrile-Pt
donation relative to the neutral Pt atom, thereby blue-shifting
the νCN frequency.
The comparison between the corresponding ∆ν(exp) values
with these ∆ν(DFT) estimates given in Tables 2 and 4,
respectively, is very informative. Most significantly, the modes
displaying moderate or large positive ∆ν(exp) values are
mirrored largely by significant or substantial ∆ν(DFT) values.
In particular, the two modes, ν6a and ν1, which exhibit the largest
∆ν(exp) values yield similarly pronounced ∆ν(DFT) estimates.
Of the remaining ring modes, only the ν7a vibration shows large
∆ν(DFT) values for the Pt-group metals, but not for gold (Table
4). This finding is in harmony with the behavior of the Pt-
benzonitrile complex (Table 3) but not the metal surfaces,
reflecting bonding differences in these coordination environ-
ments. However, the small (j10 cm-1) |∆ν(DFT)| values seen
for the remaining ring modes largely match the experimental
behavior, with the exception of the ν11 vibration.
Relation to Chemisorption-Induced Electron Polarization
and Bond-Length Changes. The observation of a broad
concordance between the mode-dependent ∆ν(exp) and ∆ν(DFT)
behavior gives us sufficient confidence in the reliability of the
DFT calculations for mimicking the effects of metal surface,
as well as metal atom, coordination to utilize this approach to
interpret the experimental wavenumber shifts in terms of
chemisorption-induced electronic polarization and intramolecular
bond-length changes. Table 5 lists the electronic charges, q,
residing on each atom of uncoordinated benzonitrile, as
estimated by DFT with the Mulliken method,34 along with the
charge changes (∆q) seen upon metal coordination for the same
Bolstered by these findings, we next intercompare the ∆ν(exp)
values with corresponding calculated frequency shifts, ∆ν(DFT),
obtained by coordinating the nitrile nitrogen to metal atoms.
Table 4 lists the latter values extracted for the vibrational modes
in Tables 2 and 3 for three 5d metals: iridium, platinum, and
gold (i.e. 5d8, d9, and d10). For the last metal, ∆ν(DFT) values
are given also for the Au3 cluster. (While DFT calculations were
also undertaken for the other metals considered here, given the
approximations involved the bonding trends are most clearly
discernible within such a Periodic row.) Two ∆ν(DFT) entries
are given for coordination to single metal atoms. The left-hand
column refers to the metal-benzonitrile complex with the
metal-N bond distances, rM-N, at their equilibrium values (as
computed by DFT), given in the column header. While the rM-N
values are intuitively reasonable, being longer for the d10 metal
atom (gold) than the Pt-group adducts, the differences (about
0.5 Å) are larger than expected for bonding to the different metal
surfaces. To accommodate this, the adjacent right-hand columns
list ∆ν(DFT) values calculated for the three metals in Table 4
with rM-N constrained at a common “average” value, 2.00 Å.
[The ∆ν(DFT) values given for the Au3 cluster refer only to
this bond length.] The sensitivity of the ∆ν(DFT) values to rM-N
is clearly evident, with their magnitude increasing markedly as
rM-N decreases for a given metal. This complication renders
the metal-dependent ∆ν(DFT) values of semiquantitative value
only; however, the observed sensitivity of the frequency shifts
to the vibrational mode is qualitatively unaffected by the choice
rM-N distances as in Table 4. (The atom numbering system used
in Table 5 is clarified in the structure shown in Figure 2.) While
estimating such “absolute” atomic charges suffers from some
arbitrariness, at least the changes in electron density upon
coordination, of primary interest here, should approximately be
represented by the ∆q values. Summarized in Table 6 are
equilibrium bond distances, r, calculated by DFT for free
benzonitrile and the changes, ∆r, induced upon coordination
to Ir, Pt, and Au atoms and to Au3, as before. The uncoordinated
r values are uniformly close to those estimated from microwave
spectra for gaseous benzonitrile.23
Inspection of Table 5 shows that, perhaps not surprisingly,
the largest coordination-induced electron-density changes occur
in the nitrile substituent atoms, with the nitrogen gaining and
the C7 carbon losing substantial (ca. 0.2 e-) charge. These
effects can be rationalized in terms of metal-nitrile π back-
of rM-N
.
Comparison between the corresponding ∆ν(DFT) values for
a single Au atom and the Au3 cluster in Table 4 (both for rM-N
) 2.00 Å) indicates a close similarity for most aromatic ring
modes, although the “substituent” vibrations ν7a and νCN are
blue-shifted significantly for only the Au trimer. These differ-
ences are unsurprising given that ν7a and especially νCN should
be most closely coupled with the metal. More significantly,
however, the Au monomer and trimer yield large, but closely
similar, blue-shifts for the ν6a and ν1 ring modes. Overall, the
above trends therefore support the semiquantitative validity of
utilizing single metal atoms in the DFT calculations as a means
of mimicking the perturbations of at least the aromatic ring
vibrations induced upon benzonitrile chemisorption.
(34) Mulliken, R. S. J. Chem. Phys. 1962, 36, 3428.

Periodic Trends in Electrode-Chemisorbate Bonding
J. Am. Chem. Soc., Vol. 123, No. 51, 2001 12823
Table 5. Coordination-Induced Changes in Mulliken Charges on Benzonitrile Atoms As Calculated by DFT
change in atomic charge,c ∆q
Ir
Pt
Au
Au3
uncoord.
atoma
charge,b q
1.80 Å
2.00 Å
1.79 Å
2.00 Å
2.28 Å
2.00 Å
2.00 Å
C1
-0.143
0.088
0.042
0.044
0.260
-0.426
0.030
-0.019
-0.018
0
-0.012
-0.023
-0.001
-0.010
+0.189
-0.230
-0.011
-0.004
-0.006
+0.147
-0.008
-0.017
0.000
-0.007
+0.216
-0.154
-0.005
-0.002
-0.003
+0.003
-0.011
-0.019
-0.001
-0.008
+0.208
-0.207
-0.008
-0.003
-0.004
+0.084
-0.007
-0.014
0.000
-0.006
+0.220
-0.140
-0.004
-0.001
-0.002
-0.027
-0.006
-0.004
+0.002
+0.002
+0.236
-0.091
+0.002
+0.004
+0.005
-0.157
-0.010
-0.008
+0.002
+0.001
+0.258
-0.181
0.000
+0.005
+0.005
-0.071
-0.012
-0.003
+0.004
+0.003
+0.275
-0.220
+0.002
+0.007
+0.009
-0.075
C2,6
C3,5
C4
C7
N
H2,6
H3,5
H4
M
a Benzonitrile carbon and hydrogen atoms are numbered (as conventional) as shown in Figure 2. b Mulliken atomic charges34 (in electron units)
as estimated from the present DFT analysis for uncoordinated benzonitrile. c Change in Mulliken atomic charge, ∆q, upon metal-benzonitrile
coordination, with rM-N values indicated in column header, such that a negative ∆q indicates a gain of e- charge and a positive ∆q represents a
loss of e- charge.
seen to be compressed upon coordination, especially to Pt and
Ir (Table 6). While the |∆r| values within the aromatic moiety
are relatively small, reflecting the minor changes in electron
density, a net ring expansion upon Pt-group metal coordination
is evident via changes in the C1-C2 and (to a lesser extent)
the C3-C4 bonds. These effects are consistent with the
corresponding increases in electron density at the C1 and C4
positions, indicating the occurrence of antibonding electron
repulsion.
Armed with the structural information in Tables 5 and 6, it
is of particular interest to examine the motion of specific carbon
atoms in the various benzonitrile “ring-centered” modes to link
the experimental coordination-induced vibrational behavior with
the manner and degree of electron polarization and bond-length
changes. (Note that, unlike the νCN mode which is largely
decoupled from other vibrations due to the high C-N force
constant, some “ring” modes will involve substantial vibrational
coupling between the aromatic ring and substituent motions.)
The ADF-DFT output provides detailed information regarding
the various atomic displacements, the salient aspects of which
are now summarized for the vibrational modes of particular
interest here. Significantly, both the ν6a (“in-plane bend”) and
ν1 (“ring breathing”) modes involve substantial C1-C7-N
stretching motions, especially the C1-C7 bond, with the ring
and substituent distortions being in-phase and out-of-phase,
respectively. The third coordination-sensitive vibration according
to DFT (and seen experimentally for the Pt-benzonitrile
complex), the ν7a (“C-X stretching”) mode, also involves C1-
C7 stretching coupled with ring motion. In contrast, the ν12 (“C-
C-C trigonal breathing”) and ν18a (“C-H in-plane bending”)
modes feature ring distortion without any C1-C7-N stretching
motion. The same situation also applies to the ν9a (“C-H in-
plane bend”) mode. The sole identifiable b1 symmetry mode,
ν11, also involves some degree of C1-C7-N stretching coupled
with bending motion. Although this mode yields small negative
∆ν(DFT) values (Table 4), larger positive ∆ν(exp) values are
obtained in both the Pt-complex and chemisorbed environments
(Tables 2,3).
Figure 2. Structure of coordinated benzonitrile, with the atom
numbering system as used in Tables 5 and 6.
donation, yielding charge residing primarily on the more
electronegative N atom, together with charge induction from
C7 to N produced by metal-nitrile σ donation. However, the
C1 carbon (i.e. that bonded to the nitrile) along with the C2
atom experience significant increases in electron density via
conjugative effects, with C4 (the para position) also gaining
charge upon coordination to Pt and Ir, but not Au. This
difference in bonding between the Pt-group and coinage metals
is also evident in the loss and gain in their electronic charge,
respectively, seen upon coordination, undoubtedly reflecting the
greater π back-donating ability of the former. Indeed, this factor
is probably also responsible for the νCN red-shifts, predicted by
DFT upon Pt-group coordination, and the νCN blue-shifts
obtained for Au3 binding (Table 4), in qualitative accordance
with the experimental metal-dependent νCN behavior (Table 2)
(see also below). Significantly, closely comparable ∆q values
were obtained for coordination to not only the Au atom and
Au3 (Table 5) but also to the Au5 cluster (not shown). This
concordance further supports the validity of the single metal-
atom DFT results for the present purposes.
Further DFT-based insight regarding coordination-induced
perturbations in benzonitrile bonding is provided by the ∆r
values in Table 6. Nitrile coordination to Pt and Ir, but not to
Au (or Au3), is seen to significantly elongate the C-N bond
(Table 6). Given the close relationship between bond vibrational
frequencies and equilibrium bond lengths, commonly known
as “Badger’s rule”,35 the observed negative ∆νCN values (i.e.,
red-shifts) observed on the Pt-group metals (Tables 2 and 3)
are to be expected. Interestingly, however, the C1-C7 bond is
Consequently, then, the presence of moderate or substantial
vibrational frequency shifts for certain “ring modes” upon nitrile-
induced benzonitrile complexation or chemisorption apparently
emanates from an involvement of C1-C7-N stretching in the
coupled motion, given the above finding of substantial coordi-
nation-induced electron density and bond-length alterations in
(35) Nakamoto, K. Infrared and Raman Spectra of Inorganic and
Coordination Compounds, 5th ed.; Wiley: New York, 1997; Part A, p 15.

12824 J. Am. Chem. Soc., Vol. 123, No. 51, 2001
Mrozek et al.
Table 6. Coordination-Induced Changes in Benzonitrile Bond Lengths, ∆r (Å), as Calculated by DFT
bond length changes,c ∆r, Å
Ir
Pt
Au
Au3
uncoord.b
bonda
r, Å
1.80 Å
2.00 Å
1.79 Å
2.00 Å
2.28 Å
2.00 Å
2.00 Å
C1-C2
C2-C3
C3-C4
C1-C7
C7-N
1.388
1.375
1.380
1.411
1.154
+0.006
-0.002
+0.001
-0.017
+0.010
+0.006
-0.001
+0.001
-0.015
+0.005
+0.005
-0.002
+0.001
-0.017
+0.006
+0.004
-0.002
+0.001
-0.015
+0.002
+0.001
-0.001
0
-0.004
-0.002
+0.002
-0.001
0
-0.007
+0.001
+0.002
-0.002
0
-0.013
-0.003
a The bond numbering system is shown in Figure 2. b The equilibrium bond distance in uncoordinated benzonitrile, as estimated by DFT. c Change
in equilibrium bond distances upon coordination to the metal atom indicated, as estimated by DFT, such that a negative ∆r indicates a shortening
and a positive ∆r represents a lengthening of the bond.
the C1-C7-N moiety (Tables 5 and 6). Most significantly,
Badger’s rule can readily account for the marked coordination-
induced blue-shifts, i.e., positive ∆ν(exp) [and ∆ν(DFT)] values,
observed especially for the ν6a and ν1 vibrations, given that the
Cl-C7 bond compresses significantly upon Pt-group coordina-
tion (Table 6). Even though the ν6a and ν1 vibrations involve
coupled ring-C1-C7-N distortion, and the nitrile νCN (“iso-
lated C-N motion”) exhibits a red-shift (for Pt and Ir) along
with C7-N bond elongation (vide supra), the dominance of the
C1-C7 motion in these coupled modes undoubtedly leads to
the net observed blue-shifts. Since small or negligible ∆ν(exp),
as well as ∆ν(DFT), values are obtained for modes featuring
purely aromatic ring motion, one may conclude that the minor
changes in electron density obtained for the C1, C2, C6, and
C4 atoms do not contribute primarily to the observed frequency
shifts, at least not for the ν6a, ν11, and ν7a modes. Nevertheless,
the small (<10 cm-1) yet detectable red-shifts, i.e., negative
that the metal d-band center moves up in energy as one
progresses from right to left along a Periodic row, thereby
encouraging back-donation. Similar trends are also seen from
recent DFT calculations for CO chemisorbed on Pt-group (111)
surfaces,38 although the experimental metal-dependent variations
in CO stretching frequencies are complicated by differences in
preferred bonding-site geometries and other bonding factors.
Given the limitations of the present DFT calculations for
predicting the ∆νCN values, alluded to above, a reliable general
analysis of these metal-dependent trends would clearly require
the use of larger cluster (or even slab) surface models; this is
beyond the scope of the present work.
Of course, however, the major virtue of the present coupled
SERS-DFT results, especially compared to IRAS data, is in our
ability to discern metal-dependent trends in vibrational frequen-
cies for numerous other intramolecular adsorbate modes. Of
particular significance in this regard are the “aromatic-ring”
modes, most prominently ν6a and ν1, that involve substantial
coupled C1-C7-N motion, especially the C1-C7 bond. The
coordination-induced blue-shifts in these modes, driven by C1-
C7 bond compression as noted above, exhibit similar trends
along a Periodic row as for the νCN red-shifts, increasing in
the order Au < Pt < Ir and Ag < Pd j Rh (Table 2). This
sequence, d10 < d9 < d8, is again captured by the corresponding
DFT-based frequency shifts (Table 4). The corresponding DFT
bond-length data (Table 6) show that this sequence is mirrored
most notably by greater C1-C7 bond compression. The nature
of these metal-dependent trends suggests strongly that the
coordination-sensitive ring-mode blue-shifts are also triggered
chiefly by metal-adsorbate back-donation, acting now to
strengthen the C1-C7 bond.
It remains to consider the metal-dependent nature of the small,
yet measurable, coordination-induced red-shifts of ring modes,
such as ν12, ν18a, ν9a, ν19a, and ν8a (Table 2), that do not involve
significant contributions from C1-C7-N motion. These trends
in SERS frequencies, also mirrored largely by the DFT
calculations (Table 4), exhibit metal-dependent red-shifts, again
in the sequence d10 < d9 < d8. Table 6 shows that these “pure
ring-mode” red-shifts correlate primarily with increases in the
C1-C2 bond distances, with smaller changes in C2-C3 and
C3-C4 bonds. Examination of the corresponding Mulliken
charges in Table 5 suggests that these mode frequency and bond-
distance changes arise from alterations in the antibonding ring
electron density in the sequence d10 < d9 < d8.
∆ν(exp) as well as ∆ν(DFT) values, observed for the ν12, ν18a
,
and ν19a vibrations (Tables 2-4) can readily be linked (via
Badger’s rule) with the net ring expansion associated with the
C1-C2 and C3-C4 bonds (Table 6), attributed to the increases
in antibonding electron density on the C1, C2, and C4 atoms
(Table 5).
While this analysis is largely descriptive, it evidently provides
a self-consistent rationale for the notable sensitivity of the metal
coordination-induced frequency shifts, in sign as well as
magnitude, to the coupled nature of normal-mode vibrations
on the basis of localized bonding perturbations.
Metal-Dependent Chemisorbate Bonding. As noted at the
onset, a central theme in this study concerns understanding the
manner in which the metal surface-adsorbate binding depends
on the electrode material. Even though we have already alluded
to this issue in the foregoing discussion, it is appropriate here
also to summarize some overall metal-dependent trends as
extracted from the present combined SERS-DFT approach. In
general, nitrile coordination to metals is expected to involve
both ligand-metal σ donation and π back-donation. Similarly
to metal carbonyls, since the latter bonding mode involves an
antibonding nitrile orbital its occurrence is expected to induce
ν
CN red-shifts.33 Examination of the metal-dependent SERS νCN
behavior (Tables 1 and 2) shows that the νCN frequencies red-
shift in the sequence Cu < Ag < Au j Pt < Pd ∼ Rh ∼ Ir.
While the experimental νCN differences are small, these trends
therefore suggest that the extent of metal-nitrile back-donation
increases from right to left in the Periodic series, i.e., in the
sequence d10 < d9 j d8. This observation [also mimicked by
DFT (Table 4)] is consistent with theoretical expectations,36,37
including those based on recent DFT calculations,36 indicating
Consequently, then, these ring-mode red-shifts, along with
the larger blue-shifts seen for coupled ring-substituent motion
and the νCN mode red-shifts, each exhibit metal-dependent trends
that are consistent with metal-adsorbate back-donation increas-
(36) Hammer, B.; Nørskov, J. K. AdV. Catal. 2000, 45, 71.
(37) (a) Sung, S.-S.; Hoffmann, R. J. Am. Chem. Soc. 1985, 107, 578.
(b) Shustorovich, E.; Baetzold, R. C.; Muetterties, E. L. J. Phys. Chem.
1983, 87, 1100.
(38) (a) Koper, M. T. M.; van Santen, R. A.; Wasileski, S. A.; Weaver,
M. J. J. Chem. Phys. 2000, 113, 4392. (b) Wasileski, S. A.; Koper, M. T.
M.; Weaver, M. J. J. Phys. Chem. B 2001, 105, 3518.

Periodic Trends in Electrode-Chemisorbate Bonding
J. Am. Chem. Soc., Vol. 123, No. 51, 2001 12825
ing from right to left along a given Periodic row (i.e. d10 < d9
< d8). The behavioral differences between the 4d and 5d series
are more subtle, although red-shifts for the “pure ring-mode”
vibrations as well as the blue-shifts for the “coupled ring-
substituent” modes appear to be more pronounced for the 5d
than for the corresponding 4d metals (Table 2).
of computational efficiency.36,39 The effect of the variable
surface potential in electrochemical systems can also be
monitored by applying external electrostatic fields.38,40 Such
field-dependent vibrational DFT calculations have so far focused
on small chemisorbates such as CO, similar to the emphasis
placed in modeling metal-UHV interfaces by DFT. Neverthe-
less, there are clearly considerable opportunities for gaining a
deeper understanding of metal-chemisorbate bonding for
molecularly more complex adsorbates, such as the present
example, by combining vibrational spectroscopy with DFT
model calculations. Further studies along these general lines
for organic as well as inorganic adsorbates at electrochemical
interfaces, emphasizing SERS and/or IRAS measurements
combined with DFT, are being pursued in our laboratory.
Concluding Remarks
The present findings are considered to provide a clear
illustration of the utility of our “overlayer SERS” strategy for
exploring metal-dependent bonding of complex organic adsor-
bates in electrochemical environments. The excellent (e1 cm-1
)
wavenumber resolution of SERS, combined with its ability to
detect a large fraction of the chemisorbate normal modes,
enables uniquely detailed vibrational information to be extracted,
even at metal-solution interfaces. The value of DFT for aiding
band assignments, as well as the fundamental interpretation of
the chemisorbate vibrational properties in terms of surface
bonding, is also clearly evident. While the present DFT
calculations, mimicking surface coordination by using a single
metal atom or small metal cluster, are somewhat crude, more
realistic surface models employing finite clusters or slabs can
in principle be utilized, albeit at the (perhaps substantial!) cost
Acknowledgment. This work is supported by the National
Science Foundation (Analytical and Surface Chemistry Program)
and by the Petroleum Research Fund, administered by the
American Chemical Society.
JA010049K
(39) For example: van Santen, R. A.; Neurock, M. Catal. ReV.-Sci. Eng.
1995, 37, 557.
(40) For example: (a) Head-Gordon, M.; Tulley, J. C. Chem. Phys. 1993,
175, 37. (b) Illas, F.; Mele, F.; Currulla, D.; Clotet, A.; Ricart, J. M.
Electrochim. Acta 1998, 44, 1213.