Gas-phase studies of group-11 cation (Cu+, Ag+, and Au+) reactions with 2-propanol in a supersonic beam-expansion source

Article abstract of DOI:10.1246/bcsj.74.677

The gas-phase reactions of group-11 ions (Cu⁺,Ag⁺, and Au⁺) with 2-propanol were investigated using laser vaporization combined with supersonic beam expansion in a time-of-flight mass spectrometer. A number of reaction pathways, including assoication, dehydrogenation, and dehydration, were observed. Among these metal ions, Au⁺ is highly reactive toward 2-propanol. The experimental results are parallel to studies by Wilkins et al. using an FT-ICR mass spectrometer (J. Am. Chem. Soc., 107, 7316 (1985)). The geometries and binding energies of all complexes M⁺-L, where M⁺ = Cu⁺, Ag⁺, and Au⁺, L = 2-propanol, acetone, propene, and H₂O, were estimated using Moller-Plesset perturbation (MP2) and density functional theory (BLYP). The bond strengths of the complex ions involving gold revealed the strongest bonding, due to a relativistic effect. The energetic relationship between the reactants and the products facilitated an interpretation of the observed exit channels. Because of the supersonic beam source employed, larger clusters containing metal ions were generated as well. Because the cluster sizes with solvent molecules were ≥ 2, the reactivities were apparently limited to Cu⁺ and Ag⁺ clusters, while the reactive nature changed in Au⁺ along with the formation of acetone, lacking in the bare Au⁺ reactions. For larger clusters, calculations on Ag⁺ - (2-propanol)₂ and Ag⁺ - (2-propanol)(acetone) complexes using BLYP were performed to obtain their energetics.

Full text of DOI:10.1246/bcsj.74.677

c. Jpn.
© 2001 The Chemical Society of Japan
Bull. Chem. Soc. Jpn., 74, 677—688 (2001)
677
e Chemical
ciety of Ja-
n
Gas-Phase Studies of Group-11 Cation (Cuꢀ, Agꢀ, and Auꢀ) Reactions
with 2-Propanol in a Supersonic Beam-Expansion Source
e Chemical
ciety of Ja-
n
Gas-Phase Studies of Group-11 Cation Reactions
Y.-C. Huang et al.
Ya-Chien Huang, Po-Hua Su, and Chen-Sheng Yeh*
01
7
8
Department of Chemistry, National Cheng Kung University, Tainan, Taiwan 701, R. O. C
(Received August 18, 2000)
SJA8
09-2673
287
gust
The gas-phase reactions of group-11 ions (Cuꢀ,Agꢀ, and Auꢀ) with 2-propanol were investigated using laser va-
porization combined with supersonic beam expansion in a time-of-flight mass spectrometer. A number of reaction path-
ways, including assoication, dehydrogenation, and dehydration, were observed. Among these metal ions, Auꢀ is highly
reactive toward 2-propanol. The experimental results are parallel to studies by Wilkins et al. using an FT-ICR mass spec-
trometer (J. Am. Chem. Soc., 107, 7316 (1985)). The geometries and binding energies of all complexes Mꢀ–L, where
Mꢀ ꢁ Cuꢀ, Agꢀ , and Auꢀ, L ꢁ 2-propanol, acetone, propene, and H2O, were estimated using Møller–Plesset perturba-
tion (MP2) and density functional theory (BLYP). The bond strengths of the complex ions involving gold revealed the
strongest bonding, due to a relativistic effect. The energetic relationship between the reactants and the products facilitat-
ed an interpretation of the observed exit channels. Because of the supersonic beam source employed, larger clusters con-
taining metal ions were generated as well. Because the cluster sizes with solvent molecules were ≥ 2, the reactivities
were apparently limited to Cuꢀ and Agꢀ clusters, while the reactive nature changed in Auꢀ along with the formation of
acetone, lacking in the bare Auꢀ reactions. For larger clusters, calculations on Agꢀ–(2-propanol)2 and Agꢀ–(2-pro-
panol)(acetone) complexes using BLYP were performed to obtain their energetics.
00
00
.2
Probing the reactions of metal ions with organic substrates
in the gas phase without any solvent interference is of particu-
lar importance, due to catalytic implications involving bond-
activating processes through the metallic centers. There is no
doubt that understanding the structures and interacting forces
between the metals and molecules will facilitate a complete
comprehension of metal ion–molecule reactions. The geome-
tries in metal ion-molecule complexes are strongly dependent
on how the metal ions interact with the coordinated molecules.
The electrostatic or covalent nature in bonding has been de-
scribed for a variety of complexes.1–6
For gas-phase metal ion-molecule chemistry, Fourier-trans-
form ion cyclotron resonance7 and guided ion-beam8 mass
spectrometers have provided significant contributions. Both
tecniques have obtained thermochemical data under ther-
molized conditions. It should be mentioned that the method
used to generate complex ion species might have a significant
influence on the final product distributions. Recently, a molec-
ular beam pick-up method as the ion–molecule reaction source
was employed to study bare metal ions or molecular cluster ion
reactions.9–11 Under those circumstances, the association and
H-elimination process switch as the cluster sizes change has
been observed in alkaline earth metal ion–water clusters.9 The
alkaline earth metal ion–methanol clusters appeared to have
similar behaviors to those in H2O systems.11 However, Dun-
can and co-workers have reported that such intracluster reac-
tions with the loss of hydrogen, which leads to
MgOCH3ꢀ(CH3OH)nꢂ1, were inhibited in a pulsed beam-ex-
pansion source.12 This was explained in terms of clusters pro-
duced at colder temperatures from their source. Therefore, the
motivation was to investigate the reactions of group-11 metals
(Cuꢀ, Agꢀ, and Auꢀ) with 2-propanol using a pulsed-beam ex-
pansion source.
Staley et al.13 and Wilkins et al.14 studied Cuꢀ, Agꢀ, and
Auꢀ with a series of alcohols (methanol, ethanol, 1-propanol,
2-propanol, 1-butanol, isobutanol, and 2-methyl-2-propanol)
using an FT-ICR mass spectrometer. The intriguing findings
were that Cuꢀ, Agꢀ, and Auꢀ exhibited distinct reactivities to-
ward these alcohols. A diversity of processes, such as dehy-
drogenation or dehydration, occurred in association with metal
insertion into the C–O or C–H bond. Wilkins proposed oxida-
tive addition via α C–H bond insertion to proceed dehydroge-
nations, as opposed to an O–H insertion mechanism, as sug-
gested by Staley.13,14 In the Feꢀ, Crꢀ, and Moꢀ reactions with
alcohols, Gross thougth that the high O–H bond energy did not
favor metal-ion insertion based on a thermodynamic consider-
ation.15 Auꢀ displayed the highest reactivity, while Agꢀ pro-
duced fewer reaction products. According to Wilkins’s results
with Agꢀ, Agꢀ–ROH condensations were always the predomi-
nant products, except for the 2-propanol and 2-methyl-2-pro-
panol reactions.14 Moreover, we also noted that the most ex-
tensive products were observed in those reactions of Auꢀ with
2-propanol and 1-butanol.14 Thus, 2-propanol was chosen as a
target molecule.
In this study, we attempted to investigate whether a beam-
expansion source, analogous to Duncan’s design,12 can pro-
duce similar reactions to those of Wilkins et al.14 In addition,
ab initio and density functional theory (DFT) methods were
employed to study the structures and binding energies of the
relevant Mꢀ–L reaction products, where Mꢀ ꢁ Cuꢀ, Agꢀ, and
Auꢀ, L ꢁ 2-propanol, acetone, propene (C3H6), and H2O. The
relative energies between the reactants and the products can

678 Bull. Chem. Soc. Jpn., 74, No. 4 (2001)
Gas-Phase Studies of Group-11 Cation Reactions
then be attained so as to examine the thermodynamic stability.
Since a supersonic molecular beam was used in our experi-
ments, the reactions of larger metal ion–molecule clusters were
also studied. Further, theoretical calculations were conducted
on the larger cluster ions, Agꢀ–(2-propanol)2 and Agꢀ–(2-pro-
panol)(acetone), to facilitate interpreting the experimental ob-
servations.
Results and Discussion
Reactions of Bare Metal Ions with 2-Propanol. Prior to
seeding 2-propanol vapor, pure He gas was used to inspect the
ion products from our ion source. Figure 1 shows that only
Cuꢀ, Agꢀ, and Auꢀ atomic signals appeared, and that no other
impurities, i.e. metal clusters and oxides, were detected. The
Cuꢀ and Agꢀ mass spectra show doublet patterns for the iso-
topes for each element. Copper has two isotopes, 63 (69.1%)
and 65 (30.9%), and silver possesses 107 (51.8%) and 109
(48.2%) with nearly equivalent abundances.
Under the same experimental conditions, e.g. vaporization
power and backing pressure, as shown in Fig. 1, ion-molecule
reactions took place whenever 2-propanol was employed in the
He gas. Figure 2a shows that several reaction products were
observed in the low-mass region (< 150 amu). The Cuꢀ reac-
tions with 2-propanol primarily yielded Cuꢀ–H2O, Cuꢀ–C3H6,
and Cuꢀ–CH3COCH3, as well as the Cuꢀ–CH3CH(OH)CH3
association channel. In addition to complexation, a peak along
with a Cuꢀ signal was weakly generated and assigned as
C3H7Oꢀ. Taking the assumption of Wilkins, C3H7Oꢀ is re-
ferred to CH3Cꢀ(OH)CH3. Acetone (mass 58) is two mass
units less than 2-propanol (mass 60). Hence, a triplet band lo-
cated at the mass 120 region is ascribed to both the Cuꢀ–ace-
Experimental
A description of the apparatus used for this experiment was
previously given.16,17 A laser-vaporization technique combined
with a supersonic molecular beam was employed to induce metal
ion-molecule reactions. A copper (99.999%) or silver (99.9%) rod
was suspended in a cutaway holder, a rod holder without a growth
channel to produce primarily cation complexes containing one and
two ligands, attached to a Jordan pulsed valve, whereas gold foil
(99.9%) was wrapped around an Al rod for producing Au ions.
The metal rods were irradiated using a 532 nm wavelength of the
second-harmonic output of a Nd:YAG laser (Quantel Brilliant)
operated at 10 Hz with 1–2 mJ/pulse. The vaporization laser beam
was focused 1.5 cm ahead of the metal rods. Helium carrier gas
with a back pressure of 6 atm seeded with 2-propanol vapor was
expanded supersonically and intersected with metal ions at the va-
porized region. HPLC-grade 2-propanol was used in these experi-
ments. CH3CH(OD)CH3 (ACROS, 98ꢀ%) and CH3CD(OH)CH3
(Aldrich, 98%) were used to clarify the reaction mechanisms.
These deuterated samples were loaded into a liquid reservoir,
where the buffer gas carried the 2-propanol vapor into a pulsed
valve, and conducted inside a glovebox filled with nitrogen gas.
The ion products were expanded and skimmed (1 mm diame-
ter) into a reflectron time-of-flight mass spectrometer. The posi-
tive ions were extracted by applying pulsed voltages to two metal
plates, i.e. repeller and extraction plates, and traveled along a
field-free tube. They were then reflected using a double-stage re-
flector and detected using a dual microchannel plate detector.
Theoretical Computations
The caluculations were carried out using second-order Møller–
Plesset perturbation theory (MP2) and the BLYP density function-
al theory. The double-zeta valence basis sets and the effective
core potentials (ECPs) of Hay and Wadt were used for Cu, Ag, and
Au atoms.18 In order to prevent an imbalance in the basis set be-
tween the metals and ligands, the split-valence double-zeta 6-31G
basis was restricted for our calculation choice to the ligands.
Three basis sets, i.e. 6-31G, 6-31G(d), and 6-31G(d,p), were se-
lected to optimize the H2O and Mꢀ–H2O complexes, where Mꢀ ꢁ
Cuꢀ, Agꢀ, and Auꢀ. The relativistic effects were included in the
Hay–Wadt ECPs of Ag and Au. We also added the f polarization
functions to increase the flexibility of the ECP basis. The f expo-
nents are 3.525 for Cu, 1.611 for Ag, and 1.050 for Au.19 Once the
basis set was chosen, the ligand and metal cation-ligand complex
geometries were obtained at the MP2 and BLYP theory levels with
6-31G(d,p) plus ECP on metals. The harmonic vibrational fre-
quencies were computed at the MP2 and BLYP levels to clarify
whether the minima were genuine, and to calculate the zero-point
energy. For metal ion–(2-propanol) complexes, the zero-point en-
ergies were calculated at the BLYP/6-31G(d,p). Natural charge
populations were studied on the complexes at the MP2 level using
a natural bond orbital (NBO) analysis. The calculations were per-
formed using a Gaussian 98 software package.20
Fig. 1.
atomic signals.
Mass spectra corresponding to Cuꢀ, Agꢀ, and Auꢀ

Y.-C. Huang et al.
Bull. Chem. Soc. Jpn., 74, No. 4 (2001) 679
Fig. 3.
Photodissociation mass spectrum of Cuꢀ–C3H6 at
355 nm with laser intensity of 1.4 mJ/pulse. The depletion
signal represents the parent ions with two copper isotopes
and is accumulated based on a data acquisition cycle; pho-
todissociation laser on subtracted from laser off.
intensities of our resulting ions can be varied to some extent by
appropriately tuning the experimental parameters, e.g. the tim-
ing between the vaporization laser and the extraction voltages.
In order to investigate any possible internal temperature effect,
He cooling gas was replaced by Ar with a larger polarization.
The resulting products from Ar were the same as those from
He. Our product channels are in good agreement with
Wilkins: the formation of H2O and C3H6 complexes via dehy-
dration, the gen eration of acetone complexes via dehydroge-
nation, and the production of CH3Cꢀ(OH)CH3 accompanied
by CuH. Figure 2a also displays larger solvent clusters con-
taining two molecules. As can be seen, copper ions still persist
with dehydrogenation and dehydration process reactivities,
whereas the reaction products diminished significantly in mag-
nitude. We discuss this issue later in the text.
Fig. 2.
Mass spectra for the reactions of 2-propanol with
(a) Cuꢀ, (b) Agꢀ, and (c) Auꢀ.
tone and Cuꢀ–(2-propanol) complexes and the middle peak,
with a mass of 123, consisting of 63Cuꢀ–(2-propanol) and
65Cuꢀ–acetone, two components. With regard to Cuꢀ–C3H6. a
photodissociation action was conducted to facilitate the ideni-
fication of C3H6 by means of their fragments. The photodisso-
ciation laser beam crossed the ion beam at the turning point of
the reflector in the RTOF mass spectrometer. The fragment
and parent ions were analyzed using their flight times. Figure
The mass spectrum in Fig. 2b illustrates that the Agꢀ–ace-
tone complex was the only detectable product in addition to
the simple Agꢀ–(2-propanol) association in the reactions of
Agꢀ with 2-propanol. The experimental conditions focused
mainly on the generation of Agꢀ coordinated with one mole-
cule adduct, as shown in Fig. 2b. From the reaction outcomes,
Agꢀ has been shown to be much less reactive toward 2-pro-
panol compared with Cuꢀ. We found the Agꢀ–(2-propanol)
complex to have a dominant abundance over that of the Agꢀ–
acetone channel. This fact is invariant to the experimental con-
ditions, which is contrary to Wilkins’ observations, where the
product distributions were 84% and 16% for Agꢀ–acetone and
Agꢀ–(2-propanol), repectively.14
3 shows the resulting fragments as Cuꢀ together with a charge-
transfer channel, C3H6
ꢀ
.
Power-dependence measurements
using 355 nm revealed a two-photon process for C3H6ꢀ forma-
tion. A parallel experiment was performed by seeding pure
propene gas into the He carrier gas to from the Cuꢀ–C3H6
complex. Identical consequences (same fragments as shown in
Fig. 3 and a two-photon absorption at 355 nm) were obtained
from photodissociation under the same experimental condi-
tions. This implies that the C3H6 species is propene, although
the other isomeric structures cannot be completely excluded.
The C3H6ꢀ observation via photodissociation indicates that an
intact propene is bound to Cuꢀ. On the basis of Wilkins’ re-
sults, the product distributions were 4% for C3H7Oꢀ, 9% for
Cuꢀ–H2O, 21% for Cuꢀ–C3H6, 21% for Cuꢀ–CH3COCH3, and
45% for Cuꢀ–CH3CH(OH)CH3.14 It should be noted that the
If the ion-melecule chemistry shifted to glod, Auꢀ induces
manifold products involving a number of dissociative attach-
ment reactions with 2-propanol, as depicted in Fig. 2c. Be-
sides the channels yielded from the dehydration process, a
strong peak regarded as CH3Cꢀ(OH)CH3 (C3H7Oꢀ) accompa-
nied by AuH and an appropriate intensity of CH3CꢀH(OH)
(C2H5Oꢀ) with the formation of AuCH3 appeared. Wilkins re-
ported the largest production (39%) in CH3Cꢀ(OH)CH3 and a
19% distribution in C2H5Oꢀ.14 We also observed a very small

680 Bull. Chem. Soc. Jpn., 74, No. 4 (2001)
Gas-Phase Studies of Group-11 Cation Reactions
Scheme 1.
tion mechanisms. Figure 4 displays the reactions of Cuꢀ with
CH3CH(OD)CH3 and CH3CD(OH)CH3 and of Agꢀ with
CH3CD(OH)CH3. Based on a careful examination, a quartet
pattern was observed in the location of the Mꢀ–CH3CD(OH)
CH3 mass regions. They were assigned as Mꢀ–CH3COCH3
and Mꢀ–CH3CD(OH)CH3 adducts including two isotopes in
metals for each complex. As expected, dehydration with com-
petitive expulsion of H2O or C3H6 appears to be the oxidative
addition of metal ions into the C–O bond, followed by β-hy-
drogen migration with a subsequent elimination of water or
propene. For the dehydrogenation pathway, this process pro-
ceeds through metal-ion insertion into either the α C–H or O–
H bond. Subsequently, an α-hydride shift is followed by re-
ductive elimination to retain acetone, forming Mꢀ–acetone.
Unfortunately, our deuterated experiments were not able to
distinguish whether the C–H or O–H bond is involved in the
metal-ion oxidative addition. However, this is most likely to
occur on the lesser binding energy of C–H, as proposed by
Wilkins14 and Gross.15 The binding energies of α C–H and O–
Fig. 4.
Mass spectra for the reactions of deuterium-substi-
tuted CH3CH(OD)CH3 with (a) Cuꢀ and CH3CD(OH)CH3
with (b) Cuꢀ and (c) Agꢀ.
H are 381.221 and 438.9 kJ molꢂ1 22 respectively. The primary
,
amount of Auꢀ–(2-propanol), while no Auꢀ–acetone could be
identified. Both 2-propanol and acetone complexes had pro-
duction yields ≤ 1% in Wilkins’ studies.14 It was quite surpris-
ing that there was such a small amount of Auꢀ–(2-propanol)
condensation, which was expected to be readily synthesized
under the beam conditions in our experiments. In Fig. 2c,
some Auꢀ free clusters, e.g. (C2H5O)(C3H6)ꢀ, (C3H7O)(C3H8–
O)ꢀ, were generated from our source.
Basically, our results are consistent with a previous investi-
gation by Wilkins and co-workers. Cuꢀ, Agꢀ, and Auꢀ prima-
rily undergo dehydration, dehydrogenation, and/or hydride ab-
straction pathways, depending on their reactivities. These
dissociative attachment reactions were proposed to result from
metal-ion insertion into the C–O, or C–H bond.14,15 The the-
modynamic factor governs the final products produced.14,15
Deuterated reagents, CH3CH(OD)CH3 and CH3CD(OH)–
CH3, were conducted to identify the above-mentioned inser-
reaction routes are depicted in Scheme 1. In addition to the de-
hydration mechanism, Scheme 1a suggests how the hydroxide
abstraction takes place in gold cation reactions, in spite of its
low production yield. Wilkins reported a 4% C3H7ꢀ formation
value.14 In contrast to hydroxide abstraction, hydride abstrac-
tion, forming AuH in Scheme 1b, occurs readily. For hydride
abstraction, copper ions do not show much reaction with 2-
propanol and Agꢀ is totally inert, as can be seen in Fig. 2.
Scheme 1c illustrates the metal ion insertion into the C–C bond
for the formation of C2H5Oꢀ in Auꢀ reactions.
The thermodynamic factor is always one of the consider-
ations in ion–molecule reactions. With the aid of the theoreti-
cal approach, the energetic relationship between reactants and
products can be estimated to facilitate the interpretation of the
Mꢀ–L complexes observed, where Mꢀ ꢁ Cuꢀ, Agꢀ, and Auꢀ,
L ꢁ 2-propanol (CH3CH(OH)CH3), acetone (CH3COCH3),

Y.-C. Huang et al.
Bull. Chem. Soc. Jpn., 74, No. 4 (2001) 681
Table 1.
as a Function of Methods and Basis Sets and Comparison with the Literature Values
Binding Energies (Do in kJ molꢂ1) of Complexes of Cuꢀ, Agꢀ, and Auꢀ with H2O and CH3COCH3
Cuꢀ–H2O
Agꢀ–H2O
Auꢀ–H2O
Agꢀ–CH3COCH3
126.8
MP2
6-31G
199.6
169.5
167.4
143.1
118.4
116.3
183.7
158.2
156.5
6-31G(d)
6-31G(d,p)
BLYP
6-31G
6-31G(d)
6-31G(d,p)
230.1
198.7
199.6
152.7
126.8
126.8
217.1
191.2
191.6
145.2
Exptl
146.4a)
139.3e)
121.8f)
159.8ꢃ18.8i)
160.7ꢃ7.5b)
Theory
162.3c)
149.0d)
123.0g)
150.2h)
160.7g)
a) Taken from reference 23. b) Taken from reference 24. c) Taken from reference 25. d) Taken from ref-
erence 26. e) Taken from reference 27. f) Taken from reference 28. g) Taken from reference 5. h) Taken
from reference 2. i) Taken from reference 29.
Table 2.
Binding Energies (Do, kcal molꢂ1) of Cuꢀ, Agꢀ, and Auꢀ Complexes Using 6-31G(d,p)
Cuꢀ
Agꢀ
Auꢀ
2-Pro-
panola)
193.3
233.9
Ace-
toneb)
190.0
243.1
C3H6
H2O
2-Pro-
panola)
134.7
145.2
Ace-
toneb)
126.8
145.2
C3H6
H2O
2-Pro-
panola)
200.8
241.0
Ace-
toneb)
198.3
256.9
C3H6
H2O
MP2
BLYP
169.0
232.2
167.4
199.6
106.7
123.4
116.3
126.8
253.6
300.0
156.5
191.6
a) CH3CH(OH)CH3. b) CH3COCH3.
propene (C3H6), and H2O. Upon these metal ion–ligand spe-
cies, H2O with the retention of Cuꢀ, Agꢀ, and Auꢀ and Agꢀ–
CH3COCH3 are the only examples where the experimental
and/or the theoretical values in the binding energy are avail-
able.23–29 In order to select an appropriate level to describe the
above complexes, we performed calculations by varying the
ligand basis sets, i.e. 6-31G, 6-31G(d), and 6-31G(d,p), on the
smallest Mꢀ–H2O adducts.
kJ molꢂ1 less strongly bound than the energy calculated by
Ma.5 Compared with BLYP, the MP2 performance is less ac-
curate in predicting the bond energy for the Agꢀ–ligand com-
plexes. For the Auꢀ–H2O system, a previous value of 150.2 kJ
molꢂ1 by Schwarz and co-workers using the CCSD(T) level of
theory is the only available data.2 It turns out that the MP2 ap-
proach gives an estimation which is only 6.3 kJ molꢂ1 larger
than the Schwarz’s result.
As shown in Table 1, the convergence of the binding ener-
gies is attained at the 6-31G(d,p) basis for both the MP2 and
BLYP methods. The value estimated by MP2 using 6-
31G(d,p) on Cuꢀ–H2O is within the margin of error of the
In addition to σ-donor molecules, such as H2O, CH3COCH3,
and CH3CH(OH)CH3, in our studies, a π-bonded ligand, e.g.
C3H6, was generated and coordinated with Cuꢀ and Auꢀ as
well. The unsaturated C3H6 molecule is a candidate to simulta-
neously act as a σ-donor and a π-acceptor. That is, the interac-
tion with metal ions involves metal ion ← C3H6 electron dona-
tion and metal ion → C3H6 π-back donation. In a survey of
previous studies in C3H6 (propene) complexing with Cuꢀ,
Agꢀ, and Auꢀ, Schröder et al. semiquantitatively predicted
measurements by Armentrout et al.24 and is 5.1 kJ molꢂ1
,
slightly more than the Bauschlicher et al. value.25 However, it
is greater than the experimental value of Michl23 and the theo-
retical calculation from Ohanessian26 by approximately of 18
kJ mol.ꢂ1 In contrast, BLYP resulted in excessive energy in
binding, and a Cs symmetry was attained at BLYP/6-31G(d,p)
level. Previous studies25,26 all exhibited a C2v symmetry for
Cuꢀ–H2O as the same structure, calculated by the MP2 meth-
od. Hence, the high binding energy and incorrect estimated
symmetry make BLYP/6-31G(d,p) useless in describing the
Cuꢀ–H2O system. With respect to the interaction of Agꢀ with
H2O, the BLYP method presented good agreement with early
reports in both theoretical and experimental measure-
ments.5,27,28 On the other side, the binding energy was slightly
underestimated by MP2. If we move on to examine Agꢀ–
CH3COCH3 using the 6-31G(d,p) basis, the BLYP prediction
resides within the limits of the experimental values29 and 15.5
Auꢀ–C3H6 with a binding energy of 292.9ꢃ29.3 kJ molꢂ1 30
.
The computations resulting from MP2 and BLYP using 6-
31G(d,p) produce 253.6 and 300.0 kJ molꢂ1 for Auꢀ–C3H6, re-
spectively. Lacking accurate experimental data, it is essential-
ly difficult to make a Do(Auꢀ–C3H6) comparison between MP2
and BLYP. Koch et al. investigated C2H4 complexes contain-
ing Cuꢀ, Agꢀ, and Auꢀ with binding energies computed as
189.9, 137.7, and 260.7 kJ molꢂ1, respectively.4 In consider-
ation of the inductive effect, a methyl group replacing a hydro-
gen atom on propene would favor electron transfer from the
ligand to the metal ions, subsequently increasing the binding
energy by the coordination of propene to Mꢀ. The issue per-

682 Bull. Chem. Soc. Jpn., 74, No. 4 (2001)
Gas-Phase Studies of Group-11 Cation Reactions
Table 3.
Natural Population Analysis of Mꢀ–L Complexes
Mꢀ ꢁ Cuꢀ
Mꢀ ꢁ Agꢀ
Mꢀ ꢁ Auꢀ
L ꢁ CH3CH(OH)CH3
L ꢁ CH3COCH3
L ꢁ C3H6
Charge of M
Populationsa)
Charge of M
Populationsa)
Charge of M
Populationsa)
Charge of M
Populationsa)
0.94
0.09
0.93
0.11
0.92
0.15
0.96
0.07
0.97
0.03
0.97
0.03
0.95
0.07
0.98
0.02
0.84
0.25
0.84
0.28
0.79
0.45
0.92
0.15
s
s
s
s
L ꢁ H2O
a) Populations refer to 4s, 5s, and 6s for Cuꢀ, Agꢀ, and Auꢀ, respectively.
Fig.
6.
Optimized
structures
for
(a)
Agꢀ–
CH3CH(OH)CH3, (b) Agꢀ–CH3COCH3, (c) Agꢀ–C3H6,
and (d) Agꢀ–H2O at MP2/6-31G(d,p) and BLYP/6-
31G(d,p) (in parenthesis). Selected bond lengths in Å and
anglis in degree.
Fig.
5.
Optimized
structures
for
(a)
Cuꢀ–
CH3CH(OH)CH3, (b) Cuꢀ–CH3COCH3, (c) Cuꢀ–C3H6,
and (d) Cuꢀ–H2O at MP2/6-31G(d,p) and BLYP/6-
31G(d,p) (in parenthesis). Selected bond lengths in Å and
angles in degree.
taining to electron donation will be discussed immediately. It
is, then, expected that the bond strength is Auꢀ–C3H6 > Auꢀ–

Y.-C. Huang et al.
Bull. Chem. Soc. Jpn., 74, No. 4 (2001) 683
C2H4.
The binding energies of Cuꢀ, Agꢀ, and Auꢀ with a series of
ligands are listed in Table 2. It was found that BLYP has eval-
uated larger binding energies compared with the MP2 calcula-
tions. With the same ligands, the trends of cation stabilization
by coordination are in the order Agꢀ < Cuꢀ < Auꢀ. Cuꢀ–H2O
was computed to be slightly larger than Auꢀ–H2O in binding
energy at either MP2 or BLYP. An increase in Do upon coordi-
nation with Auꢀ has been interpreted in terms of relativistic ef-
fects.4 Relativity causes a significant contraction of the Auꢀ
(s0d10) valence 6s orbital, while this effect is minor for Cuꢀ and
Agꢀ. Consequently, the electrophilic character of the 6s orbit-
al is enhanced, which results in a significant electron transfer
from the ligand to Auꢀ. Thus, this procedure increases the co-
valent bonding between Auꢀ and the ligands. Table 3 gives re-
sults of the natural population analysis of metal cation–ligand
complexes. As expected, the 6s populations of Auꢀ are dis-
tinctly promoted, suggesting a marked ligand → Mꢀ σ-dona-
tion. For Cuꢀ and Agꢀ complexes, the Cuꢀ–lignad adducts ex-
hibit more effective electron transfer. With respect to the
valence d orbital populations, a survey of each valence d orbit-
al is doubly occupied with population > 1.96 for Cuꢀ and Agꢀ,
while a lowering of the population, on the averafe, to 1.9 for
the highest filled d orbials of Auꢀ in the respective complex is
Our computational analysis seems to direct the use of MP2/
6-31G(d,p) in estimating the binding energies of the σ-donor
molecules complexing with Cuꢀ and Auꢀ, whereas BLYP/6-
31G(d,p) is more reliable for the π-bonded C3H6 complexes.
Interestingly, coordination with Agꢀ appears to involve a sim-
plified use of BLYP/6-31G(d,p) with various ligands. Our pro-
cedure has revealed the complexity of choosing one uniform
level of theory to analyze the transition-metal complexes. In
conclusion, we employed 6-31G(d,p) in both the MP2 and
BLYP methods to model the target complexes. Qualitatively,
they are sufficient to describe the interactions between metal
ions and molecules, the geometries of the complexes and the
relative energetics between the reactants and products.
observed. This indicates a negligible Mꢀ → ligand (Mꢀ
ꢁ
Cuꢀ and Agꢀ) back donation, though only a small contribution
involved the Auꢀ electron donation. With regard to the group-
11 metal cations interplaying with ligands, it is clear that this
interaction is predominantly an electrostatic force, as can be
seen from the metal charges listed in Table 3; nevertheless, the
participation of covalency arising mainly from ligand → Mꢀ
electron donation is important in Auꢀ–ligand bonding.
The relativistic effect also affects the complex geometries,
such as the bond lengths and bond angles. Figures 5–7 depict
the structural features. Regarding the Mꢀ–C3H6 complexes,
Bauschlicher and co-workers studied two different four-mem-
bered rings involving the formation of Mꢀ–C bonds responsi-
ble for Mꢀ–C3H6, where Mꢀ ꢁ Scꢀ, Yꢀ, and Laꢀ.31 Those
two four-membered rings include one allyl-like structure with
one hydrogen bound to the metal and the other with a hydro-
gen attached to the C3H6ꢀ central carbon. On the basis of our
photodissociation, resulting in an intact C3H6, we have at-
tempted to exclude the allyl-like structure. That with a hydro-
gen bound to the central carbon was chosen as a model, as well
as the configurations showm in Figs. 5–7. It appears that the
former structures are approximately 62.8 kJ molꢂ1 less stable
in energy than the geometries given in Figs. 5–7.
Optimization at MP2 and BLYP clearly display that the
Mꢀ–ligand bond distance decreases in the order Agꢀ > Auꢀ >
Cuꢀ in complexation of the same ligand. A purely electrostatic
interaction would have the bond-to-ligand distance increased
from Cuꢀ to Auꢀ. In an inspection of 2-propanol complexes,
∠HOMꢀ follows the order Agꢀ > Cuꢀ > Auꢀ. Such decreased
bond angles represent metal ions bound toward a lone pair of
electrons of an oxygen atom, which lies on the molecular
plane. The other lone pair is perpendicular to the molecular
plane in 2-propanol. An analogous phenomenon was observed
at ∠MꢀOC for acetone complexes, where one oxygen lone
pair lies on the molecular plane perpendicular to the acetone
dipole moment. The deviation from C2v symmetry, as the Cs
Fig.
7.
Optimized
structures
for
(a)
Auꢀ–
CH3CH(OH)CH3, (b) Auꢀ–CH3COCH3, (c) Auꢀ–C3H6,
and (d) Auꢀ–H2O at MP2/6-31G(d,p) and BLYP/6-
31G(d,p) (in parenthesis). Selected bond lengths in Å and
anglis in degree.

684 Bull. Chem. Soc. Jpn., 74, No. 4 (2001)
Gas-Phase Studies of Group-11 Cation Reactions
Fig. 8.
Sechmatic energy profile calculated at MP2/6-31G(d,p) for the product channels in the reactions of (a)
CuꢀꢀCH3CH(OH)CH3, (b) AgꢀꢀCH3CH(OH)CH3, and (c) AuꢀꢀCH3CH(OH)CH3. E is relative energy.
structure in Auꢀ–H2O, demonstrates the covalent contribution
in the Auꢀ–H2O moiety with Auꢀ pointing toward one of the
lone pairs. The dominant interactions are ion-dipole forces re-
sulting in a C2v symmetry for both the Cuꢀ–H2O and Agꢀ–H2O
complexes. A nonplanar Cs structure for Auꢀ–H2O was com-
puted by Schwarz et al.2
In addition to the observation of marked covalency in the in-
teractions of the σ-donor ligands with Auꢀ, a drastically en-
hanced binding energy appears in the π-bonded Auꢀ–C3H6
system, as showm in Table 2. The smaller difference in the
ionization potential between the gold cation and the ligands
would promote such an electron-transfer process. Both IP val-
ues are 9.73 and 9.22 eV for C3H6 and Au, respectively.21 The
small energy gap, i.e. ∆IP ꢁ 0.51 eV, suggests the preference
of electron donation from C3H6 to Auꢀ, as evident in Table 3.
The electron-transfer behavior reflected its effect on the CꢁC
bond length. Figures 5–7 illustrate that the CꢁC bond length-
ens in the order as Agꢀ < Cuꢀ < Auꢀ, ranging from 1.36 Å to
1.41 Å (MP2) or 1.38 Å to 1.43 Å (BLYP).
Following the bonding configurations in those metal ion–
molecule complexes, the energetic information on the complex
ions provides a guide to interpret the dehydration, dehydroge-
nation, and association pathways. Figurs 8 and 9 depict the en-
ergy changes of the metal complexes from the reactants, if one
takes into account the MP 2 and BLYP calculations, respec-
tively. Both MP2 and BLYP reveal consistent trends in de-
scribing the Cuꢀ and Agꢀ stabilization by coordination, even
though, the relative quantitative energies from both methods
are poorly matched. As can be seen in Figs. 8a–b and 9a–b,
the reaction paths in dissociative attachment and direct con-

Y.-C. Huang et al.
Bull. Chem. Soc. Jpn., 74, No. 4 (2001) 685
Fig. 9.
Schematic energy profile calculated at BLYP/6-31G(d,p) for the product channels in the reactions of (a)
CuꢀꢀCH3CH(OH)CH3, (b) AgꢀꢀCH3CH(OH)CH3, and (c) AuꢀꢀCH3CH(OH)CH3. E is relative energy.
densation are, indeed, exothermic. This explains, at least, the
formation of 2-propanol, acetone, H2O, and propene complex-
es with Cuꢀ and the generation of 2-propanol and acetone
complexes containing Agꢀ. However, the absence of Agꢀ–
H2O and Agꢀ–C3H6 in these experiments cannot be simply ex-
plicated by the thermodynamic picture. An additional poten-
tial-energy surface calculation is helpful to present a distinct
explanation. Experimentally, the hydride abstraction produced
a neutral metal hydride and C3H7Oꢀ in the Cuꢀ and Auꢀ reac-
tions with 2-propanol. Hence, the products relative to the reac-
tants in energies were also calculated for both Cuꢀ and Agꢀ re-
actions. These computations lead to the formation of CuH and
C3H7Oꢀ lacated above the reactants at MP2, but below the dis-
sociation limit of Cuꢀ–(2-propanol) at BLYP. However, AgH
and C3H7Oꢀ appear to be endothermic channels for both meth-
ods. An inspection of the calculated binding energies of CuH
(MP2:203.3 kJ molꢂ1, BLYP:269.9 kJ molꢂ1) and AgH
(MP2:162.8 kJ molꢂ1, BLYP:220.9 kJ molꢂ1), BLYP results
give a good agreement with the experimental reports
(CuH:278.2 kJ molꢂ1, AgH:215.5ꢃ7.9 kJ molꢂ1).21 In con-
sideration of the binding energies of the π-bonded C3H6 com-
plexes mentioned earlier, BLYP seems to present a better de-

686 Bull. Chem. Soc. Jpn., 74, No. 4 (2001)
Gas-Phase Studies of Group-11 Cation Reactions
scription of the covalent nature bonding under the level of the
method employed here. Qualitatively, the exit channels, i.e.
metal hydride and C3H7Oꢀ, have the highest energies among
these products. We observed C3H7Oꢀ weakly in Cuꢀ reactions
with no sign of hydride abstraction for Agꢀ in these experi-
ments.
In the case of Auꢀ, the sequence of the energy products is
quite different from the MP2 and BLYP performance. The
greatest deviation appears in the Auꢀ–(2-propanol) moiety
from the BLYP results, which do not give rise to the lowest
surface energy for the condensation process. The association
complex is expected to be produced without the existence of a
barrier and to have the minimum energy in the entire reaction
pathway. If we take account of the little yield of Auꢀ–(2-pro-
panol) in our supersonic source, it shows that Auꢀ has mani-
fested an intriguing behavior in the reactions. As stated earlier,
Auꢀ appears to be highly reactive toward 2-propanol with the
involvement of a number of oxidative-addition and reductive-
elimination mechanisms, leading to the generation of Auꢀ–
H2O, Auꢀ–C3H6, AuH, C3H7Oꢀ, Au–CH3, and C2H5Oꢀ. On
the basis of Figs. 8c and 9c, these processes are accessible for
the reactions of Auꢀ with 2-propanol under thermal conditions.
However, it is still not known whether there is no appearance
of the dehydrogenation product Auꢀ–acetone in the experi-
ments. For example, the loss of H2 is supposed to be a more
thermodynamically favored route than the loss of C3H6, fol-
lowed by the formation of Auꢀ–H2O. Therefore, kinetic con-
trol may, again, play a crucial role. Finally, concerning
Do(AuH), the prediction (290.8 kJ molꢂ1) of BLYP is in con-
cert with the experimental value (292.5ꢃ7.9 kJ molꢂ1)21, while
it is 42.7 kJ molꢂ1 less strongly bound from a MP2 evaluation.
In this study, laser ablation was introduced to generate metal
ions. It is known that laser-ablated metal ions can be formed in
electronically excited states and readily access the endother-
mic pathways relative to the ground-state reactions. Although
we cannot exclude the excited-state ions involved in the reac-
tions, the reaction products are consistent with the results from
Wilkins et al,14 using an FTICR mass spectrometer. In particu-
lar, we also have seen no production of Agꢀ–C3H6, Agꢀ–H2O
and Auꢀ–CH3COCH3 channels from our source. Otherwice,
such chemical specificity would likely be washed out in the ex-
cited-state reactions. The same inference can be drawn con-
cerning the reactions of Agꢀ–(2-propanol)n clusters, as shown
later. It is possible that the observed ion-molecule reactions
were mainly from the ground-state reactivity.
Fig. 10.
Mass spectra of the larger clusters containing (a)
Cuꢀ, (b) Agꢀ with Agꢀ–{CH3CH(OH)CH3}n as the domi-
nant products, and (c) Auꢀ.
ducts. Different from Cuꢀ and Agꢀ, Auꢀ changes the reactivi-
ty drastically with the formation of acetone channels, as shown
in Figure 10c. In the reactions of bare Auꢀ with 2-propanol,
no Auꢀ–acetone complex was observed. Here, acetone was
readily produced for the cluster size n ≥ 2. No Auꢀ–(2-pro-
panol)n association products were generated. The condensa-
tion mechanism is difficult due to the extremely high activity
of Auꢀ. It should be noted that the cluster-ion reactions can be
classified into the following mechanisms: (1) subsequent clus-
ter growth after a bare metal ion–molecule reaction, (2) a clus-
ter ion–molecule reaction following cluster formation, and (3)
an intracluster metastable reaction. 32 According to the results
given in Figure 10, reaction path (1) may be excluded. It is ex-
pected to show all cluster sizes containing H2O and C3H6
ligands if process (1) proceeds. Either reaction (2) or (3) can
contribute to the reaction products observed in our experi-
ments. However, the nature of both processes has a substantial
difference. For instance, although the changes in the metal re-
activities, as evident in Figure 10, are related to the solvation
effect for reaction (2), it is explained by the cage effect if reac-
tion (3) occurs. At this stage, we are unable to distinguish both
processes in our experimental configuration.
Metal Ion–(2-Propanol) Clusters Reactions. As men-
tioned earlier concerning the reactions of Cuꢀ with 2-propanol,
the product intensities diminished rapidly once the solvent
molecules increased. Similarly, both Agꢀ and Auꢀ revealed
reactivity change when the second solvent started to coordinate
with the metal centers. Figure 10a displays that there was no
formation of Cuꢀ complexes containing C3H6O, C3H6, and
H2O with cluster sizes ≥ 3 in the reactions of Cuꢀ with 2-pro-
panol. Figure 10b shows an example in Agꢀ–(2-propanol)n
clusters. Clusters of up to eight 2-propanol molecules were
produced. At n ꢁ 1, it is clear that the dehydrogenation chan-
nel Agꢀ–acetone is alone with the Agꢀ–(2-propanol). Based
on a careful examination, however, we observed no reaction
product with a cluster size ≥ 2, except for the association ad-

Y.-C. Huang et al.
Bull. Chem. Soc. Jpn., 74, No. 4 (2001) 687
Fig. 11.
Optimized structures for (a) Agꢀ–{CH3CH(OH)CH3}2 and (b) Agꢀ–{CH3CH(OH)CH3}(CH3COCH3) at BLYP/6-
31G(d,p). (c) The energy profile calculated at BLYP/6-31G(d,p) for the product channels in the reactions of Agꢀ–
CH3CH(OH)CH3ꢀCH3CH(OH)CH3. Selected bond lengths in Å and angles in degree. E ꢁ relative energy.
For larger clusters with n ≥ 2, theoretical calculations cho-
sen on Agꢀ–(2-propanol)2 and Agꢀ–(2-propanol)(acetone), as-
suming a dehydrogenation process, were performed. Based on
a previous analysis, BLYP/6-31G(d,p) was used to model these
clusters. The geometries were fully optimized. Figure 11a and
b depict the optimization configurations corresponding to
Agꢀ–(2-propanol)2 and Agꢀ–(2-propanol)(acetone), respec-
tively. The O–Agꢀ–O reveals nearly 180˚ as a linear structure.
Bauschlicher and co-workers have computed a linear Cuꢀ–
(H2O)2, resulting from sd hybridization on metal to reduce the
electron density along the metal-ligand bonding axis.25 With
respect to Agꢀ–(2-propanol)2, two ∠COH arrangements are in
a staggered configuration. The calculated binding energy (Do)
with respect to Agꢀ and two 2-propanol molecules is 266.9 kJ
molꢂ1. Although the Agꢀ–O bond distances are very similar in
Agꢀ–(2-propanol) and Agꢀ–(2-propanol)2, the binding energy
of the second CH3CH(OH)CH3 ligand coordinated to Agꢀ is
less by 20.9 kJ molꢂ1 than the first ligand coordination energy.
For Agꢀ–(2-propanol)(acetone) species, the binding energy is
found to be 268.6 kJ molꢂ1. As can be seen in Figire 11c, both
Agꢀ–(2-propanol)2 condensation and Agꢀ–(2-propanol)(ace-
tone) dehydration are the thermodynamically favored paths,
yet the condensation channel, i.e. Agꢀ–(2-propanol)2, has low-
er energy. This is consistent with an experimental observation,
as shown in Figure 10b. However, the energetics between the
reactants and the products are unable to resolve what driving
force forms acetone readily once the solvent molecules in-
crease in Auꢀ. This remains an open question.
Summary
1. The reactions of group-11 cations (Cuꢀ, Agꢀ, and Auꢀ)
with 2-propanol have been investigated by a technique using
laser vaporization combined with a supersonic beam expan-
sion. Metal ions underwent several types of mechanisms with
the involvement of C–O, C–H and/or C–C insertion. The ob-
served reaction products were parallel to the results by Wilkins
et al. using an FT-ICR mass spectrometer. Compared with
Cuꢀ and Agꢀ, the enriched chemistry reveals the high reactivi-

688 Bull. Chem. Soc. Jpn., 74, No. 4 (2001)
Gas-Phase Studies of Group-11 Cation Reactions
ty of the Auꢀ center. In contrast with the dominant association
channels produced in both Cuꢀ and Agꢀ, the Auꢀ–(2-pro-
panol) complex was yielded weakly under supersonic expan-
sion circumstances.
175 (1991).
M. Sanekata, F. Misaizu, K. Fuke, S. Iwata, and K.
Hashimoto, J. Am. Chem. Soc., 117, 747 (1995).
10 J. K. Gibson, J. Phys. Chem., 100, 15688 (1996).
11 W. Lu and S. Yang, J. Phys. Chem. A, 102, 825 (1998).
12 M. R. France, S. H. Pullins, and M. A. Duncan, Chem.
Phys., 239, 447 (1998).
13 R. W. Jones and R. H. Staley, J. Phys. Chem., 86, 1669
(1982).
14 D. A. Weil and C. L. Wilkins, J. Am. Chem. Soc., 107, 7316
(1985).
15 S. Huang, R. W. Holman, and M. L. Gross, Organometal-
lics, 5, 1857 (1986).
16 Y. S. Yang and C. S. Yeh, Chem. Phys. Lett., 305, 395
(1999).
9
2. MP2 and DFT methods were carried out on all complex-
es. The BLYP level of theory evaluated stronger binding ener-
gies than the MP2 calculations. We found that it is difficult to
use one uniform level of theory to model complexes consisting
of σ-donor and π-bonded ligands. Calculations at higher lev-
els using better basis sets in both metals and ligands would
lead to a more precise description of the binding energies.
Qualitatively, both methods predicted similar trends in the
binding energies and the interactions of the metal ions with
ligands. Due to the relativistic effect, Auꢀ exhibits the stron-
gest interaction with a series of ligands. Most of the reaction
products generated can be interpreted in terms of the thermo-
dynamic stability based on the energetic relation between the
reactants and the products. The absence of Agꢀ–H2O, Agꢀ-
C3H6 and Auꢀ–acetone, however, is likely in association with
the kinetic factor.
17 Y. S. Yang, W. Y. Hsu, H. F. Lee, Y. C. Huang, C. S. Yeh,
and C. H. Hu, J. Phys. Chem. A., 103, 11287 (1999).
18 P. J. Hay and W. R. Wadt, J. Chem. Phys., 82, 299 (1985).
19 A. W. Ehlers, M. Böhme, S. Dapprich, A. Gobbi, A.
Höllwarth, V. Jonas, K. F. Köhler, R. Stegmann, A. Veldkamp, and
G. Frenking, Chem. Phys. Lett., 208, 111 (1993).
3. In addition to Mꢀ–ligand adducts, larger cluster ions,
Mꢀ–(ligand)n, were investigated as well. These reactions were
apparently confined to Cuꢀ and Agꢀ with cluster sizes ≥ 2,
while the reactivity changed, leading to the formation of ace-
tone, which was absent in the reactions of bare Auꢀ with 2-
propanol, appeared in Auꢀ–(solvent)n with n ≥ 2. Agꢀ–(2-pro-
panol)2 and Agꢀ–(2-propanol)(acetone) clusters were chosen
for theoretical computations using BLYP/6-31G(d,p). It is evi-
dent that the Agꢀ–(2-propanol)2 channel has a lower energy
than that of Agꢀ–(2-propanol)(acetone).
20 M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria,
M. A. Robb, J. R. Cheeseman, V. G. Zakrzewski, J. A. Montgom-
ery, R. E. Stratmann, J. C. Burant, S. Dapprich, J. M. Millam, A.
D. Daniels, K. N. Kudin, M. C. Strain, O. Farkas, J. Tomas, V.
Barone, M. Cossi, R. Cammi, B. Mennucci, C. Pomelli, C. Ada-
mo, S. Clifford, J. Ochterski, G. A. Petersson, P. Y. Ayala, G. Cui,
K. Morokuma, D. K. Malick, A. D. Rabuck, K. Raghavachari, J.
B. Foresman, J. Cioslowski, J. V. Ortiz, B. B. Stefanov, G. Liu, A.
Liashenko, P. Piskorz, I. Komaromi, R. Gomperts, R. L. Martin,
D. J. Fox, T. Keith, Al-Laham, C. Y. Peng, A. Nanayakkara, C.
Gonzalez, M. Challacombe, P. M. W. Dill,; B. Johnson, W. Chen,
M. W. Wong, J. L. Andres, C. Gonzalez, M. Head-Gordan, E. S.
Replogle, J. A. Pople, Gaussian 98, Revision A. 3, Gaussian Inc.,
Pittsburgh, PA (1998).
This research was supported by the National Science Coun-
cil of the Republic of China. We appreciate helpful discus-
sions with Dr. Ching-Han Hu concerning the theoretical calcu-
lations and grateful for his permission to use his computer
facilities. Generous allocations of computational time at the
National Center for High-Performance Computing of the Na-
tional Science Council are also acknowledged.
21 D. R. Lide, “CRC Handbook of Chemistry and Physics,”
79th ed, CRC Press, London (1998).
22 This is calculated from Do(RO–H•)
ꢁ
∆Hf(H•)ꢀ
∆Hf(RO•)21ꢂ ∆Hf(ROH)21 where ∆Hf(H•) ꢁ 1/2∆Hrxn(H2
→
H•ꢀH•)21 and R ꢁ CH3CHCH3
23 T. F. Magnera, D. E. David, and J. Michl, J. Am. Chem.
Soc., 111, 4100 (1989).
24 N. F. Dalleska, K. Honma, L. S. Sunderlin, and P. B.
Armentrout, J. Am. Chem. Soc., 116, 3519 (1994).
25 C. W. Bauschlicher, S. R. Langhoff, and H. Partridge, J.
Chem. Phys., 94, 2068 (1991).
References
1
C. W. Bauschicher and H. Partridge, J. Phys. Chem., 95,
3946 (1991).
2
J. Hrus˘ák, D. Schröder, and H. Schwarz, Chem. Phys. Lett.,
26 S. Hoyau and G. Ohanessian, Chem. Phys. Lett., 280, 266
(1997).
27 P. M. Holland and A. W. Castleman, J. Chem. Phys., 76,
4195 (1982).
225, 416 (1994).
3
(1996).
4
E. Magnusson and N. W. Moriarty, Inorg. Chem., 35, 5711
R. H. Hertwig, W. Koch, D. Schröder, H. Schwarz, J.
28 P. K. Chattaraj and P. v. R. Schleyer, J. Am. Chem. Soc.,
116, 1067 (1994).
29 Y. P. Ho, Y. C. Yang, S. J. Klippenstein, and R. C. Dunbar,
J. Phys. Chem. A, 101, 3338 (1997).
Hrus˘ák, and P. Schwerdtfeger, J. Phys. Chem., 100, 12253 (1996).
5
6
N. L. Ma, Chem. Phys. Lett., 297, 230 (1998).
C. N. Yang and S. J. Klippenstein, J. Phys. Chem. A, 103,
1094 (1999).
30 D. Schröder, H. Schwarz, J. Hrus˘ák, and P. Pyykkö, Inorg.
Chem., 37, 624 (1998).
31 a) M. Rosi and C. W. Bauschlicher, Chem. Phys. Lett., 166,
189 (1990). b) C. W. Bauschlicher and S. R. Langhoff, J. Phys.
Chem., 95, 2278 (1991).
7
See, for example: a) S. J. A. Martinho and J. L.
Beauchamp, Chem. Rev., 90, 629 (1990). b) B. S. Freiser, Acc.
Chem. Res., 27, 353 (1994). c) D. Schröder and H. Schwarz, An-
gew. Chem., Int. Ed. Engl., 34, 1973 (1995).
8
See, for example: a) P. B. Armentrout and R. Georgiadis,
32 We acknowledge one reviewer to mention this point.
Polyhedron, 7, 1573 (1988). b) P. B. Armentrout, Science, 251,