Spectroscopy and dynamics of methylamine. I. Rotational and vibrational structures of CH3NH2 and CH3ND2 in A states

Article abstract of DOI:10.1063/1.1575734

Rovibrational structures of methylamines in predissociative A states were investigated using (1+1) resonant-enhanced two-photon ionization (R2PI) spectroscopy. An internal rotor Hamiltonian including a torsional barrier was employed for the perfect match of experiment and theory. Ab initio results for structures and vibrational frequencies of methylamines at excited states were compared with the experiment.

Full text of DOI:10.1063/1.1575734

Spectroscopy and dynamics of methylamine. I. Rotational and vibrational
structures of CH3NH2 and CH3ND2 in states
Sun Jong Baek, Kyo-Won Choi, Young S. Choi, and Sang Kyu Kim
Citation: J. Chem. Phys. 118, 11026 (2003); doi: 10.1063/1.1575734
View online: http://dx.doi.org/10.1063/1.1575734
View Table of Contents: http://jcp.aip.org/resource/1/JCPSA6/v118/i24
Published by the AIP Publishing LLC.
Additional information on J. Chem. Phys.
Journal Homepage: http://jcp.aip.org/
Journal Information: http://jcp.aip.org/about/about_the_journal
Top downloads: http://jcp.aip.org/features/most_downloaded
Information for Authors: http://jcp.aip.org/authors
Downloaded 01 Oct 2013 to 131.104.62.10. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jcp.aip.org/about/rights_and_permissions

JOURNAL OF CHEMICAL PHYSICS
VOLUME 118, NUMBER 24
22 JUNE 2003
Spectroscopy and dynamics of methylamine. I. Rotational and vibrational
˜
structures of CH3NH2 and CH3ND2 in A states
Sun Jong Baek, Kyo-Won Choi, and Young S. Choi
Department of Chemistry, Inha University, Incheon (402-751), Republic of Korea
Sang Kyu Kima)
Department of Chemistry, Inha University, Incheon (402-751), Republic of Korea and Department of
Chemistry and School of Molecular Sciences, Korea Advanced Institute of Science and Technology (KAIST),
Daejeon (305-701), Republic of Korea
͑Received 16 December 2002; accepted 27 March 2003͒
˜
Rovibrational structures of methylamines (CH3NH2 and CH3ND2) in predissociative A states
(3s–n) are investigated using ͑1ϩ1͒ resonant-enhanced two-photon ionization ͑R2PI͒
spectroscopy. A part of experimental results was briefly reported earlier ͓J. Chem. Phys. 117, 10057
͑2002͔͒, and full detailed results and analyses are given here. Spectral origins are determined to be
41 669 and 42 038 cmϪ1 for CH3NH2 and CH3ND2 , respectively. Amino wagging and CH3 rocking
modes are optically active, giving their respective fundamental frequencies of 636 ͑487͒ and 1008
͑1012͒ cmϪ1 for CH3NH2 (CH3ND2). The CH3 moiety is found to rotate nearly freely about the
C–N axis with respect to the amino group with an accurately determined torsional barrier of
5.0Ϯ0.5 cmϪ1 at the zero-point level of CH3ND2(A). The torsional barrier increases to 19.0Ϯ0.5
˜
cmϪ1 at the (ND -wag)ϭ1 level due to wagging-torsional mode coupling. Both internal and
v
2
overall rotational fine structures are clearly resolved for the first few vibrational levels of
˜
CH3ND2(A), providing accurate values of vibrational frequencies and associated internal and
overall rotational constants. Broad spectral features of the CH3NH2 excitation spectrum are
unambiguously assigned by using the internal rotor Hamiltonian established in the analysis of the
CH3ND2 excitation spectrum. Linewidths of spectral bands provide lifetimes of corresponding
quantum states excited at particular rovibrational levels, giving, for example, ϳ8.8 and ϳ1.8 ps for
˜
zero-point and (ND -wag)ϭ2 levels of the CH ND (A) state, respectively. The lifetime of
v
2
3
2
˜
CH3NH2(A) is estimated to be much shorter, giving ␶ϳ0.38 ps at the origin band. The large H/D
isotope effect in lifetimes of excited states indicates that the primary dissociation channel is the
N–H͑D͒ bond dissociation and it proceeds via tunneling through a reaction barrier. Lifetimes are
found to be mode specific, showing the experimental fact that energy deposition to a certain
vibrational mode, which is perpendicular to the reaction coordinate, may modify the reaction barrier
along the N–H͑D͒ reaction coordinate. Ab initio results for structures and vibrational frequencies of
methylamines at excited states are compared with the experiment. © 2003 American Institute of
Physics. ͓DOI: 10.1063/1.1575734͔
group4 measured translational energies of different products
I. INTRODUCTION
from several reaction channels of the CH3NH2 dissociation
at 222 nm to conclude that the major channel is N–H bond
dissociation while the C–H and C–N bond dissociation
channels are also significant. On the other hand, the Ashfold
group,5,6 through H-atom Rydberg-tagged high-resolution
translational spectroscopy, had reached a conclusion that the
H fragment, slow and fast, is almost entirely from the N–H
bond dissociation, suggesting that the C–H bond breakage
practically does not occur. According to theoretical calcula-
tions by the Morokuma group8 for the N–H bond fission, the
first-excited state of methylamine has a 3s Rydberg character
in the Franck–Condon region and it is immediately followed
by coupling with a repulsive valence state correlating dia-
batically to the ground products. Similar to the N–H bond
dissociation of ammonia,9–15 the dissociative excited state
crosses with the ground electronic state which diabatically
The spectroscopy and dynamics of primary amines are
quite meaningful to investigate, since fundamental properties
regarding the structures and reactivity of those molecules are
essential in understanding the role of primary amines in
many chemical reactions important in organic syntheses and
biological processes. In 1963, Michael and Noyes1 reported
that the N–H bond dissociation is a dominant reaction chan-
nel with a yield of more than 75% from the first-excited
methylamine dissociation, while other reactions such as C–H
bond dissociation and H2 formation channels are ϳ7.5% and
less than 10%, respectively. More recently, sophisticated ex-
periments on the photodissociation dynamics of CH3NH2
were carried out by several research groups.2–7 The Butler
a͒
Author to whom correspondence should be addressed. Electronic mail:
˜
skkim@inha.ac.kr
correlates to excited products, HϩCH3NH(A), in the late
0021-9606/2003/118(24)/11026/14/$20.00
11026
© 2003 American Institute of Physics
Downloaded 01 Oct 2013 to 131.104.62.10. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jcp.aip.org/about/rights_and_permissions

J. Chem. Phys., Vol. 118, No. 24, 22 June 2003
Spectroscopy of methylamine. I
11027
stage along the N–H reaction coordinate. Even though these
earlier studies2–7 provide very important insights into poten-
tial energy surfaces on which many different reactions take
place, state-to-state reaction dynamic studies, which are ex-
tremely useful, had not been plausible especially since initial
reactant states have been poorly characterized until very re-
cently.
Methylamine is the simplest primary amine, and thus its
rovibrational structure has been thoroughly studied for the
ground electronic state using microwave and/or far-infrared
spectroscopy.16–23 However, for the first-excited state of me-
thylamine, assignments for the spectral origin and associated
vibrational bands had been in dispute for quite a long time.
Absorption spectra of methylamine in the gas phase at room
temperature had been taken by a number of groups.24–29 The
structure of the CH3ND2 origin absorption band taken in the
gas phase at room temperature, in which they also concluded
the internal free rotation of CH3 with respect to ND2 with a
torsional barrier less than 1 cmϪ1. In our earlier report, a
Hamiltonian for free internal rotation was also quite useful in
interpreting the CH3NH2 spectrum nicely, giving confirma-
tive assignments for vibational bands in the excitation
spectrum.31 Accordingly, spectral assignments were firmly
established giving the 0–0 transition energy of 41 669 cmϪ1
for the first-excited CH3NH2 .31 Optically active modes are
NH2 wagging and CH3 rocking modes with their respective
fundamental frequencies of 636 and 1008 cmϪ1. It is inter-
esting to note that accurately determined spectroscopic val-
ues are closer to those reported 65 years ago than more re-
cently reported values.
¨
first UV spectrum of methylamine was reported by Foster
In this work, full detailed analyses of excitation spectra
of CH3NH2 and CH3ND2 are given. Much richer informa-
tion about the excited methylamine structure is presented
here compared to the earlier brief publication.31 An internal-
rotor Hamiltonian including a torsional barrier is employed
for the perfect match of experiment and theory. Torsional
barriers are thus both precisely and accurately determined for
the first few vibrational bands in the CH3ND2 excitation
spectrum. Vibrational frequencies and associated internal and
overall rotational constants are accurately determined to give
a quantitative description of potential energy surfaces of
and Jungers in 1937.24 They determined the 0–0 transition
energy to be 41 680 cmϪ1 and found that two vibrational
modes with frequencies of 650 and 1000 cmϪ1 are optically
active, of which the former is associated with the amino
moiety while the latter is more likely associated with the
methyl group. In the more extended work reported in 1953
by Tannenbaum et al.,25 the 0–0 transition was reported to
be 41 690 cmϪ1 and vibrational bands with fundamental fre-
quencies of 660 and 1000 cmϪ1 were claimed to be due to
N–H bending and C–N stretching, respectively. Later,
Tsuboi et al.26,27 located the origin of CH3NH2 at 41 715
cmϪ1 and assigned 651 and 1004 cmϪ1 bands as NH2 wag-
ging and CH3 rocking modes, respectively. More recently,
Taylor et al.28,29 reported the first mass-resolved excitation
spectra of supersonically cooled methylamine and its isoto-
pomers. Comparing their experimental results with ab initio
calculation, Taylor et al. concluded that the spectral origin is
not observed in the experiment due to a severe geometrical
change occurring upon electronic excitation. Accordingly,
they claimed that the CH3NH2 origin is expected to be lo-
cated at 39 770 cmϪ1 and optically active modes are amino
wagging and scissoring with frequencies of 642 and 1476
cmϪ1, respectively.28 The disagreement continued in a me-
thylamine UV–VUV high-resolution absorption study re-
ported by Hubin-Franskin et al. in 2002.30 According to
them, the CH3NH2 origin is at 41 771 cmϪ1 and bands lo-
cated at 637, 984, and 371 cmϪ1 above the origin are NH2
wagging, C–N stretching, and two quanta of NH2 torsion,
respectively. These confusing disagreements, which contin-
ued for the past 65 years among previous experimental
works for excited methylamine spectroscopy, are partly re-
lated to broad spectral features of the CH3NH2 excitation
spectrum.
˜
CH3NH2 and CH3ND2 in their A states. Lifetimes are de-
duced for all vibronic bands, giving energy- and mode-
dependent predissociation rates in excited states of CH3NH2
and CH3ND2 . A plausible predissociation mechanism is also
discussed in comparison with previously reported dynamics
studies.
II. EXPERIMENT
The methylamine sample (CH3NH2 , Aldrich͒ was pur-
chased and used without further purification. As described in
the earlier publication,31 CH3ND2 was synthesized by re-
peating the process of the CH3NH2 dissolution in D2O fol-
lowed by dehydration in KOH solution. Methylamine
samples were introduced into gas cylinders to prepare 2% of
methylamine gases (CH3NH2 or CH3ND2) in neon gas. The
gas mixture was then expanded through a 0.3-mm-diam
nozzle orifice ͑General Valve͒ into a differentially pumped
molecular beam chamber equipped with two turbo-molecular
pumps. The backing pressure was ϳ3.5 atm, and the repeti-
tion rate was 10 Hz. The molecular beam was skimmed
through a 1-mm-diam skimmer ͑Precision Instrument Ser-
vices͒. Background pressures of the source and ionization
chambers were maintained at ϳ10Ϫ5 and ϳ10Ϫ7 Torr, re-
spectively, when the nozzle was on. The 355-nm output of a
Nd:YAG laser ͑Spectra-Physics, GCR-150, 10 Hz͒ was used
to pump a dye laser ͑Lambda-Physik, Scanmate II͒ to gener-
ate laser pulses in the 440–480 nm range. The time duration
of the laser pulse was ϳ6 ns, and the pulse energy was
10–20 mJ/pulse. The laser output was frequency doubled in
a BBO crystal placed on a home made autotracking system
to maintain both intensity and direction of the doubled laser
output while its wavelength was scanned. Tunable UV laser
Very recently, our group has reported the ͑1ϩ1͒
resonant-enhanced two-photon ionization ͑R2PI͒ spectra of
jet-cooled CH3NH2 and CH3ND2 , finding that spectral
broadening is actually due to the ultrashort lifetimes of pre-
dissociating A states. Because of the relatively longer life-
times of CH3ND2 excited states, its rotational fine structures
could be revealed. And it was found that the CH3 group
nearly freely rotates with respect to the amino moiety in the
excited state. It should be noted that in 1982 Tsuboi and
Yirakawa27 reported a preliminary analysis of the rotational
31
˜
Downloaded 01 Oct 2013 to 131.104.62.10. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jcp.aip.org/about/rights_and_permissions

11028
J. Chem. Phys., Vol. 118, No. 24, 22 June 2003
Baek et al.
pulses with a 0.25 cmϪ1 spectral bandwidth were generated
in the 220–240 nm region. The UV laser pulse was spatially
and temporally overlapped with the molecular beam.
electronic state of CH3NH2 (CH3ND2) where ⌫elϭA1Ј , rovi-
brational states with even Ka (⌫rotϭA1Ј ,A2Ј) and vibrational
states of AЈ and E combine with para ͑ortho͒ states. Mean-
Ј
1
Molecular ions generated via the ͑1ϩ1͒ R2PI process
were repelled, accelerated, drifted along the field-free region,
and detected by microchannel plates ͑MCP, Jordan͒ to give
the R2PI signal. The ion signal was digitized by an oscillo-
scope ͑LeCroy 9361͒ and stored in a PC, which was also
used to control the dye laser and autotracker. The ion signal
was monitored as a function of excitation wavelength to give
the excitation spectra of methylamines. Absolute wave-
lengths of the laser output were calibrated using the optogal-
vanic effect from a hollow-cathode lamp ͑Ne͒ within Ϯ0.5
while, rovibrational states with AЈ and E vibrational sym-
Ј
1
metry species of CH3NH2 (CH3ND2) with odd Ka (⌫rot
ϭA1Љ ,A2Љ) are to be combined with ortho ͑para͒ states. Simi-
larly, rovibrational states with AЉ and E vibrational symme-
Љ
2
try species of CH3NH2 (CH3ND2) with even Ka (⌫rot
ϭAЈ1 ,A2Ј) or odd Ka (⌫rotϭA1Љ ,A2Љ) give ortho ͑para͒ or
para ͑ortho͒ states, respectively. Nuclear spin statistical
weights for ortho:para states are 3:1 and 2:1 for CH3NH2
and CH3ND2 , respectively.
The 3s–n transition of methylamine is the allowed
electric–dipole transition. Analyses of rotational structures of
the CH3ND2 excitation spectrum indicate for sure that the
rotational transition is b or c type ͑vide infra͒, indicating that
the transition dipole moment is perpendicular to the C–N
axis. Since methylamine is the nonrigid molecule and espe-
cially since internal rotation is almost free in the excited
state, the symmetry of electronic state alone may not be the
only criterion for the optical selection rule. In that case, se-
lection rules are determined by considering the symmetry of
the direct product of electronic (⌫ele) and internal-rotor
(⌫int) symmetry species. Considering nuclear spin conserva-
tion in the electronic transition (⌬Iϭ0), ground states of
cmϪ1
.
III. RESULTS AND DISCUSSION
The electronic transition from the ground electronic state
to the lowest excited state of methylamine occurs by the
excitation of a nitrogen lone-pair electron to a 3s Rydberg
orbital. The excited state, however, is predissociative even at
the origin, indicating that the Rydberg state is strongly
coupled to repulsive valence states leading to decomposition
of the molecule via N–H, C–N, or C–H bond fission.8
Therefore, the transition of methylamine to the first-excited
state is a bound-to-metastable transition, which carries not
only spectroscopic information of the Franck–Condon re-
gion, but also dynamic information involved in quantum-
specific predissociation processes. In this sense, the R2PI
spectra of CH3NH2 and CH3ND2 provide detailed informa-
tion about the structure and dynamics of those molecules in
the excited states. Since lifetime broadening is much less
int
which ⌫ele ꢀ ⌫ ϭAЈ , AЉ , E , and E are symmetry al-
Љ
Љ
Ј
Љ
1
2
lowed to excited states with ⌫ele ꢀ ⌫ ϭAЉ , AЈ , E , and
int
Ј
Ј
Љ
2
1
E
(AЉ , AЈ , E , and E ), respectively, for the c-type ͑or
Љ Ј
1 2
Ј
b-type͒ transition. However, in the Franck–Condon transi-
tion, within the Born–Oppenheimer approximation, the elec-
tronic symmetry of the 3s Rydberg state is most likely to be
AЉ , which correlates to A or A for planar or ciscoid con-
Ј
Љ
2
˜
severe in CH3ND2(A), peaks representing rotational struc-
formers in the Cs ͑MS͒ group, respectively. In this case,
symmetry species of the internal-rotor states are important
criteria for determining selection rules. In other words, for
example, in the c-type transition, electronically excited
tures are nicely resolved in the CH3ND2 excitation spectrum.
Therefore, spectral analyses for CH3ND2 are presented here
prior to those for CH3NH2 . The D-to-H substitution effect is
then applied for analyzing the CH3NH2 excitation spectrum.
This section is presented in the following order: ͑A͒ internal
rotation with symmetry consideration, ͑B͒ CH3ND2 , ͑C͒
CH3NH2 , and ͑D͒ tunneling and predissociation.
internal-rotor states of AЈ , AЉ , E , and E symmetry spe-
Ј
Љ
1
2
cies are only symmetry allowed.
A Hamiltonian for the internal and overall rotation in-
cluding a torsional barrier for methylamines in the excited
state can be written as
A. Internal rotation and selection rules
Methylamine is a nonrigid molecule, and thus complete
nuclear permutation inversion ͑CNPI͒ group theory should
be applied for the classification of symmetry species of quan-
tum states.32 Methylamines in ground and excited states, if
tunneling is considered, belong to the G12 group. In the
ground state, due to tunneling through a sixfold potential for
internal rotation, the zero-point energy level is split into six
states. Since the torsional barrier in the ground state is rela-
tively high (Vϳ690 cmϪ1),19 all six states in the ground
vibrational level are split within the energy much less than
0.5 cmϪ1. Accordingly, in the molecular beam, all of these
states are expected to be populated. Vibrational symmetry
species of these split states are AЈ , AЉ , E , and E , where
HϭH ϩF pϪP͒2ϩV ␾͒,
͑1͒
͑2͒
͑
͑
rot
V ␾͒ϭ V /2͒ 1Ϫcos 6␾͒ .
͑
͑
͓
͑
͔
6
Here Hrot is the standard rigid-rotor symmetric-top approxi-
mated rotational Hamiltonian and F is the effective rotational
constant for the internal rotation of the top about the C–N
axis. Since the difference of rotational constants B and C is
less than 0.05 cmϪ1 21
the symmetric-top approximation is
,
good enough within the experimental resolution of this work.
The V6 is the barrier height of a sixfold torsional potential
energy surface along the torsional angle ␾. The (pϪP) term
represents the relative angular momentum of the top and
Ј
Љ
1
2
E and E states are doubly degenerated. The symmetry spe-
Ј
Љ
frame. The free-rotor basis set
͉
m ϭ(2␲)Ϫ1/2 exp(Ϫim␾)
cies of the total wave functions of CH3NH2 and CH3ND2 are
A1Љ and A1Ј , respectively, since hydrogen nuclei are fermions
while deuterium nuclei are bosons. Therefore, in the ground
͕ ͖
͘
is used to construct the Hamiltonian matrix. Nonvanishing
matrix elements are as follows:
Downloaded 01 Oct 2013 to 131.104.62.10. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jcp.aip.org/about/rights_and_permissions

J. Chem. Phys., Vol. 118, No. 24, 22 June 2003
Spectroscopy of methylamine. I
11029
FIG. 1. R2PI excitation spectrum of CH3ND2 in the 41 500–44 200 cmϪ1 range. Vibrational bands due to ␯ (CH3 rock͒, ␯ (ND2 wag͒, and their
7
9
combination modes are appropriately assigned. The peak position expected for the next lower ND2-wagging band to the spectral origin is indicated as an
arrow, which confirms the origin assignment.
2
B. CH3ND2
m
m
͉
͉
H
H
͉
͉
m ϭ ATϩAF͒m2ϩBJ Jϩ1͒ϩ AFϪB͒K
͑ ͑ ͑
Ј
͘
͗
͗
Ϫ2AFmK ϩ V /2͒,
͑3͒
͑4͒
The CH3ND2 excitation spectrum in the 41 500–44 200
cmϪ1 region is shown in Fig. 1 with appropriate vibrational
assignments. In Fig. 1, ␯ and ␯ represent CH3 rocking and
͑
Ј
6
7
9
mϮ6 ϭϪ V /4͒.
͑
͘
6
amino-wagging modes, respectively. Combination bands of
␯ and ␯ modes are also clearly identified. Even tiny little
bands are not observed at energies below the assigned spec-
tral origin, giving confirmative evidence for the validity of
Here AT and AF are rotational constants associated with mo-
ments of inertia for the top (Itop) and frame (Iframe), respec-
tively. In principle, when the internal rotational axis is along
the C–N bond, IϭItopϩIframe or, equivalently, (1/A)
ϭ(1/AT)ϩ(1/AF). A 40ϫ40 matrix is diagonalized to give
eigenvalues and associated eigenvectors. Symmetry species
of internal-rotor wave functions can be classified from eigen-
vectors deduced from the matrix diagonalization. For ex-
7
9
ample, in the simplest case of K ϭ0 where there is no cou-
Ј
pling between overall and internal rotations, the symmetry
species of the internal-rotor wave functions which are pro-
portional to ͉0͘, ͉Ϯ1͘, ͉Ϯ2͘, ͓͉ϩ3͘ϩ͉Ϫ3͔͘, ͓͉ϩ3͘Ϫ͉Ϫ3͔͘,
͉Ϯ4͘, and ͉Ϯ5͘ are AЈ , E , E , AЉ , AЉ , E , and E ,
Љ
Ј
Ј
Љ
1
2
1
respectively.32 It should be noted that, because of the off-
diagonal matrix element in Eq. ͑4͒, free-rotor basis functions
of mϭϩ3 and mϭϪ3 couple together to generate
(1/2)1/2 Ϫ3 ] or (1/2)1/2
͉ϩ3 ϩ͉ ͉ϩ3 Ϫ͉Ϫ3 ] states of
͓ ͓
͘ ͘ ͘ ͘
which the symmetry species are AЉ2 or A1Љ , respectively. As
mentioned earlier, the symmetry selection rule indicates that
only AЉ , AЈ , E , and E (AЉ , AЈ , E , and E ) states are
Љ
Ј
Љ
Ј
2
1
1
2
FIG. 2. ͑a͒ Excitation spectrum of CH3ND2 for the origin band and ͑b͒
simulation using parameters listed in Table II. Peaks connected by dotted
lines are those very sensitive to the torsional barrier height ͑see the text for
detail͒.
symmetry allowed in the c-type ͑b-type͒ transition. Thus the
upper state of AЉ1 or A2Љ is symmetry forbidden for the c-type
or b-type transition, respectively.
Downloaded 01 Oct 2013 to 131.104.62.10. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jcp.aip.org/about/rights_and_permissions

11030
J. Chem. Phys., Vol. 118, No. 24, 22 June 2003
Baek et al.
the origin assignment. The first two bands at the origin and
ground-state methylamine over (J ,KЉ ,KЉ) quantum num-
Љ
a c
(ND -wag)ϭ1 level consist of many sharply resolved
bers is calculated from the Boltzmann distribution.
v
2
peaks, while the bands above ϳ43 000 cmϪ1 show broad
Crucial information about the structure and dynamics of
excited methylamine is also obtained from other parameters
used in the simulation. Parameters used in simulations are
listed in Table II with their uncertainties. The rotational tem-
perature of the molecular beam is found to be 3.6 K, indicat-
ing that cooling of the sample is quite sufficient. Actually, it
is found that Ne is the most effective carrier gas in cooling
methylamines in a supersonic jet compared to Ar or He. The
exact position for the 0–0 transition is precisely determined
to be 42 038Ϯ1 cmϪ1 in the simulation. It is also noteworthy
that the zero-point energy due to the torsional mode is 2.49
cmϪ1, in Table II. This small zero-point torsional energy is
important because it varies upon coupling with other vibra-
tional modes ͑vide infra͒. Predissociation dynamics are also
revealed in the spectral bandwidths. The linewidth of indi-
vidual peaks is much broader than the laser linewidth of 0.25
cmϪ1. Thus a Gaussian function with a full width at half
maximum ͑FWHM͒ of 0.25 cmϪ1 is convoluted with a
Lorentzian function with an adjustable FWHM, ⌬E, to
simulate detailed spectral features. The most successful
simulation is obtained when ⌬Eϳ0.6 cmϪ1, indicating that
the corresponding excited state lifetime ͑␶͒ is, from ⌬E ␶
ϳប, about 8.8 ps.
spectral features. This indicates that the excited-state lifetime
starts to decrease extremely sharply at the (ND -wag)ϭ2
v
2
level. This is also consistent with the intensity of the R2PI
signal, which starts to decrease quite rapidly at the same
˜
energy level, Fig. 1. That is, even though the A–X absorp-
tion cross section increases as the excitation energy increases
in the 42 000–44 000 cmϪ1 range,26,30 the ionization cross
˜
section from intermediate A states decreases in the same en-
ergy region. This indicates that the ionization process which
˜
occurs competitively with predissociation in the A state be-
comes much less efficient at vibrational levels where predis-
sociation becomes quite rapid, resulting in a decrease of the
R2PI signal. It is interesting to note that the intensity de-
creasing rate is much faster for ND2 wagging progression
bands compared to that for the CH3 rocking or its combina-
tion bands, Fig. 1.
The origin band of the CH3ND2 excitation spectrum
consists of relatively sharp peaks, Fig. 2͑a͒. The simulation
based on the overall and internal rotor Hamiltonian with a
torsional barrier ͑Sec. III A͒ is carried out for comparison
with experiment. The free-rotor energy expression used in
the simulation in our previous report was not perfectly suc-
cessful in matching all peak positions.31 However, by intro-
ducing a torsional barrier height of 5.0Ϯ0.5 cmϪ1, all experi-
mental peak positions are now perfectly reproduced in the
simulation as shown in Fig. 2͑b͒. The internal rotational con-
stants of the top (AT) and frame (AF) are both determined to
be 4.93 cmϪ1. At the zero-point energy level of the excited
state, the moments of inertia of the CH3 and ND2 group are
The band structure of the first ND2 wagging mode is
very similar to that of the origin band, Fig. 3͑a͒. The simu-
lation matches the experiment perfectly in Fig. 3͑b͒. How-
ever, the parameters used in the simulation are different from
those used for the origin band, Table II. First of all, the
torsional barrier height is found to be increased to 19.0Ϯ0.5
cmϪ1 when internal rotation is combined with one quantum
in the ND2 wagging mode. This is an interesting vibrational
dependence of the torsional barrier height, as shown in Fig.
4, providing essential information about the coupling of the
ND2 wag and internal rotation. It would be intriguing to
then found to be identical giving I (CH )ϭI (ND )
Ј
Ј
3
2
ϭ3.42 amu Å2. It is quite interesting that the moments of
inertia determined in the experiment are identical for the top
(ND2) and frame (CH3). Coupling between internal rota-
tions of the top and frame may exist especially when the
associated moments of inertia are so similar to each other. A
theoretical formulation for describing this interesting behav-
˜
compare torsional dynamics of CH3ND2(A) with those of
the CH3CH2(X) radical, since the former with one of non-
bonding electrons excited to the Rydberg state is expected to
have a rather similar electronic structure to that of the latter
in the ground electronic state. Moreover, the methylamine
cation is isoelectroic to the ethyl radical.34 Torsional barrier
heights of the ethyl radical in the ground electronic state had
been determined to be 17 and 10 cmϪ1 for the zero-point and
CH2 wag levels, respectively,35,36 giving good agreement in
general magnitude with values of barrier heights determined
ior is needed in the future. The overall rotational constant B
Ј
of the excited CH3ND2 in the symmetric-top approximation
is found to be 0.67 cmϪ1. For the ground state of CH3ND2 ,
rotational energies are calculated using an asymmetric rotor
program.33 Observed and calculated frequencies associated
with transitions between angular momentum quantum num-
bers of the lower (J ,KЉ ,KЉ) and upper (J ,K ,m ) states
Љ
Ј
Ј
Ј
a
c
˜
are listed in Table I. Energy differences between experimen-
tal and simulated values are all less than 0.5 cmϪ1. It should
be noted here that a series of peaks due to the excited
internal-rotor state with a AЉ1 symmetry and wave function of
for CH3ND2(A) here. However, the vibrational dependence
of the torsional barrier height in the ethyl radical seems to be
in the opposite trend compared to that found in
35,36
˜
˜
CH3ND2(A).
This difference of methylamine(A) and the
(1/2)1/2
͉ϩ3 Ϫ͉Ϫ3 ] is absent in the experiment. Accord-
͓
͘ ͘
ethyl(X) radical in the vibrational dependence of the tor-
sional barrier height is interesting and subjected to further
investigation.
ingly, the corresponding level is excluded in the simulation.
˜
This strongly indicates that the A–X transition belong to the
c-type transition, since in the c-type transition the AЉ1 upper
state is not symmetry allowed ͑vide supra͒. Rotational selec-
The zero-point torsional energy is 9.37 cmϪ1. Thus the
1–0 fundamental transition frequency of 487.0 cmϪ1 for
ND2 wag should be slightly modified to 480.1 cmϪ1 when
the difference between zero-point torsional energies of the
ground vibrational level and first wagging band is extracted:
i.e., 487.0ϩ͑2.49Ϫ9.37͒ϭ480.1. However, couplings among
tion rules of ⌬K (K ϪK )ϭϮ1, ⌬Jϭ0 or Ϯ1 used in the
Ј
a
simulation are also consistent with the above. Relative band
intensities associated with different upper m values are ar-
Ј
bitrarily scaled in the simulation, while the population of the
Downloaded 01 Oct 2013 to 131.104.62.10. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jcp.aip.org/about/rights_and_permissions

J. Chem. Phys., Vol. 118, No. 24, 22 June 2003
Spectroscopy of methylamine. I
11031
TABLE I. Observed and calculated frequencies with angular momentum quantum numbers of ground and
excited states for the origin bands in the 42 000–42 200 cmϪ1 region of the CH3ND2 excitation spectrum.
Downloaded 01 Oct 2013 to 131.104.62.10. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jcp.aip.org/about/rights_and_permissions

11032
J. Chem. Phys., Vol. 118, No. 24, 22 June 2003
Baek et al.
TABLE I. ͑Continued.͒
many other vibrational modes are not quantitatively known,
and in the more strict sense, such coupling effects are to be
all considered in determining fundamental frequency of pure
ND2 wag. Accordingly, in order to be fair, the fundamental
frequency of ND2 wag is reported here to be 487 cmϪ1. Both
AT and AF are found to be decreased to 4.86 cmϪ1 in the first
periment. This means that, for instance, the D–N–D angle is
actually larger than the ab initio value of 116.5°.8 Namely, if
the ab initio value of 1.020 Å is taken for the N–D bond
length in the excited state,8 then the experimental value of AeT
predicts a D–N–D angle of 129°, suggesting that better un-
derstanding of the excited-state geometry of methylamine
would benefit from more advanced theoretical calculations.
ND2 wagging band, indicating that I (CH )ϭI (ND )
Ј
Ј
3
2
ϭ3.47 amu Å2. This indicates that angles and bond lengths
become more relaxed at the first wagging vibrational level
compared to those at the origin. Applying a simple formula
AT( )ϭATϪ␣ ( ϩ 1) or AF( )ϭAFϪ␣ ( ϩ 1), it is found
The overall rotational constant B is also slightly decreased
Ј
to 0.66 cmϪ1. The individual linewidth is identical to that in
the origin band, indicating that the excited-state lifetime of
the first wagging band remains at ϳ8.8 ps.
v
v
v
v
e
e
2
2
e
e
that AeTϭAeFϭ4.97 cmϪ1 while ␣eϭ0.07 cmϪ1. Correspond-
ing values calculated from the ab initio geometry ͑CASSCF
The most dramatic change in vibrational bands due to
2␯ and ␯ is the much more broadened linewidths com-
9
7
**
with a 6-31ϩϩG ͒ of the excited CH3ND2 calculated by
pared to those in the origin and ␯ bands, Fig. 5͑a͒. Accord-
9
the Morokuma group8 are I (CH )ϭ3.16 and I (ND )
ingly, in the simulation for the 2␯ band in Fig. 5͑b͒, it is
Ј
Ј
3
2
9
ϭ3.01 amu Å2. Moments of inertia for both of the ND2 and
found that ⌬Eϳ3.0 cmϪ1, corresponding to ␶ϳ1.8 ps. Be-
CH3 groups shows relatively poor agreement with the ex-
cause of broadened linewidths, spectral features due to over-
TABLE II. Parameters used in the simulation for CH3ND2 excitation spectra.a
000
␯
2␯
␯
3␯
␯7ϩ␯
4␯
␯7ϩ2␯
2␯
7
9
9
7
9
9
9
9
AT
AF
A
4.93Ϯ0.05
4.93Ϯ0.05
2.465Ϯ0.05
0.67Ϯ0.01
0.60Ϯ0.05
8.8Ϯ0.7
4.86Ϯ0.05
4.86Ϯ0.05
2.43Ϯ0.05
0.66Ϯ0.01
0.60Ϯ0.05
8.8Ϯ0.7
͑4.77͒
͑4.77͒
͑4.5͒
͑4.5͒
͑4.1͒
͑4.1͒
͑3.9͒
͑3.9͒
͑4.6͒
͑4.6͒
͑4.6͒
͑4.6͒
͑4.1͒
͑4.1͒
B
͑0.65͒
͑0.65͒
͑0.64͒
8Ϯ1
0.7Ϯ0.1
͑25͒
͑12.24͒
͑0.68͒
2.0Ϯ1.0
2.6Ϯ0.8
͑25͒
͑12.22͒
͑0.58͒
͑8.0͒
͑0.7͒
͑0.58͒
͑8.0͒
͑0.7͒
͑0.50͒
Ͼ8.0
Ͻ0.7
͑25͒
͑12.24͒
Ј
FWHM
␶ ͑ps͒
V6
3.0Ϯ0.5
1.8Ϯ0.3
22Ϯ1
1.3Ϯ0.5
4.1Ϯ0.5
͑9.0͒
5.0Ϯ0.5
2.49Ϯ0.25
19.0Ϯ0.5
9.37Ϯ0.25
b
ZPE ͑torr͒
10.8Ϯ0.5
͑4.47͒
Evib
0
487Ϯ1
988Ϯ1
1012Ϯ1
1490Ϯ2
1522Ϯ2
1989Ϯ2
1997Ϯ2
2042Ϯ2
͓42038Ϯ1͔
aParameters in parentheses are less reliable.
bZero-point energy due to torsional motion.
Downloaded 01 Oct 2013 to 131.104.62.10. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jcp.aip.org/about/rights_and_permissions

J. Chem. Phys., Vol. 118, No. 24, 22 June 2003
Spectroscopy of methylamine. I
11033
FIG. 3. ͑a͒ Excitation spectrum of CH3ND2 for the ␯ (ND2 wag͒ band and
FIG. 5. ͑a͒ Excitation spectrum of CH3ND2 for the 2␯ and ␯ bands and
9
9
7
͑b͒ simulation using parameters listed in Table II. Peaks connected by dotted
lines are those very sensitive to the torsional barrier height, which is 19
cmϪ1 in this band.
the simulation for ͑b͒ 2␯ and ͑c͒ ␯ bands using parameters listed in Table
9
7
II. The simulation for a series of peaks at ϳ1110 cmϪ1 is shown in ͑d͒. It
should be noted that linewidths are quite different for different modes.
all rotation are not resolved, and Bϭ0.65 cmϪ1 is simply
completely conclusive since associated rotational structures
assumed in the simulation. The vibrational energy of 2␯ is
9
are not clearly resolved in the 2␯ band. Meanwhile, the
9
determined to be 988 cmϪ1, Table II. If the anharmonicity
strong peak at the vibrational energy of 1012 cmϪ1 ascribed
to ␯ (CH3 rocking͒ requires a linewidth of ϳ1.3 cmϪ1 for
constant due to the ␯ mode, ␹99 , is only considered to be
9
7
significant, then using
a simple anharmonic formula
the simulation in Fig. 5͑c͒, giving an associated lifetime of
Evib( )ϭ␻ ( ϩ 1)ϩ␹ ( ϩ 1)2 together with fundamental
v
v
v
99
e
2
2
ϳ4.1 ps. The longer lifetime of the ␯ band compared to that
7
and first overtone frequencies of 487 and 987.5 cmϪ1, re-
of 2␯ is also consistent with the higher intensity of the
9
spectively, it is found out that ␻eϭ473 cmϪ1 and ␹
99
former compared to that of the latter. This is an interesting
mode dependence of the predissociation rate, which will be
discussed more in Sec. III D. It is also interesting to note that
a series of peaks at the rovibrational energy of ϳ1110 cmϪ1
above the origin is partly rotationally resolved, Fig. 5͑a͒.
Assignments of these peaks are not certain at the present
time. However, in order to reproduce these partly resolved
rotational structures, a linewidth of ⌬Eϳ1.2 cmϪ1 has to be
used in the simulation, Fig. 5͑d͒. The corresponding lifetime
ϭ7.0 cmϪ1. The positive value of ␹ in the excited planar
99
CH3ND2 is as expected since two ground-state potentials
separated by a wagging barrier become merged in the excited
state, which is similarly observed in the inversion mode of
the NH3 excited state.9 The best fitting value of AT(2) or
AF(2) is found to be 4.77 cmϪ1 in the 2␯ band, and this
9
value is very close to 4.79 cmϪ1, which is calculated from
ATe , AFe , and ␣e values deduced from accurately and pre-
cisely determined internal rotational constants AT(0),
AF(0), AT(1), and AF(1) ͑vide supra͒. The torsional barrier
is ϳ4.4 ps, and this is much longer than the 2␯ lifetime,
9
while it is very similar to the ␯ lifetime. Actually, in the
7
height in the 2␯ band is determined to be 22Ϯ1 cmϪ1 in the
following paper,34 these peaks are found to have ␯ mode
9
7
simulation, although it should be noted that this value is not
character.
FIG. 4. Model potential V(␾)ϭ(V /2) 1Ϫcos(6␾) , for the internal rota-
͓
͔
6
tion of ND2 with respect to CH3 along the C–N axis. The V6 of 5 cmϪ1 at
the origin becomes 19 cmϪ1 at the ␯ band. The planar geometry of
FIG. 6. ͑a͒ Excitation spectrum of CH3ND2 for the 3␯ and ␯7ϩ␯ bands
9
9
9
CH3ND2 in the excited state is assumed to be more stable according to the
and the simulation for ͑b͒ 3␯ and ͑c͒ ␯7ϩ␯ bands using parameters listed
9
9
ab initio calculation in Ref. 8.
in Table II.
Downloaded 01 Oct 2013 to 131.104.62.10. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jcp.aip.org/about/rights_and_permissions

11034
J. Chem. Phys., Vol. 118, No. 24, 22 June 2003
Baek et al.
FIG. 7. ͑a͒ Excitation spectrum of CH3ND2 for the 4␯9 , ␯7ϩ2␯9 , and 2␯
7
bands and the simulation for ͑b͒ 4␯9 , ͑c͒ ␯7ϩ2␯9 , and ͑d͒ 2␯ bands using
parameters listed in Table II.
FIG. 9. R2PI excitation spectrum of CH3NH2 in the 41 500–44 000 cmϪ1
7
range. Vibrational bands due to ␯ (CH3 rock͒, ␯ (NH2 wag͒, and their
7
9
combination modes are appropriately assigned.
Vibrational bands due to 3␯ and ␯7ϩ␯ bands are
9
9
shown in Fig. 6͑a͒. Simulations for 3␯ and ␯7ϩ␯ bands
based on the parameters listed in Table II show good agree-
ment with the experiment in Figs. 6͑b͒ and 6͑c͒, respectively.
9
9
torsional barrier height cannot be deduced accurately, while
it is true that it varies with different combinations of vibra-
tional modes. Accordingly, a quantitative description of the
anharmonic coupling between ␯ and ␯ modes seems to
The vibrational term value of 3␯ is determined to be 1490
9
7
9
cmϪ1. Calculation of vibrational energy using value of ␻
e
need more spectroscopic data.
ϭ473 cmϪ1 and ␹99ϭ7.0 cmϪ1, deduced from precisely de-
termined term values of ␯ and 2␯9 , gives a 3␯ vibrational
The broad peaks found at an internal energy of ϳ2000
cmϪ1 are due to 4␯9 , ␯7ϩ2␯9 , and 2␯ bands, Fig. 7͑a͒.
9
9
7
energy of 1503 cmϪ1 ͑vide supra͒, which is very close to the
Overall, most of bands are diffuse in nature, indicating ul-
trashort lifetimes of excited states in this energy region. As
shown in Fig. 1, the band intensity at this energy is relatively
very low, though the corresponding absorption cross section
should be most intense at such internal energies. Molecular
constants deduced from simulations for these low-intensity
broadbands are thus less reliable. The lower bound for a
linewidth of ϳ8 cmϪ1 is used for the simulation for 4␯9 ,
above experimental value of 1490 cmϪ1. The linewidth in the
simulation for 3␯ is ϳ8.0 cmϪ1, corresponding to ␶ϳ0.7 ps,
9
indicating that the lifetime gets even shorter when the vibra-
tional energy is given as three quanta of ND2 wag. Mean-
while, a linewidth of ϳ2 cmϪ1 ͑␶ϳ2.6 ps͒ seems to match
the experiment pretty well for the ␯7ϩ␯ band, indicating
9
once again that the lifetime gets longer when a part or whole
of the vibrational energy is deposited onto the CH3 rocking
␯7ϩ2␯9 , and 2␯ bands as shown in Figs. 7͑b͒, 7͑c͒, and
7
mode. The vibrational term value for ␯7ϩ␯ is determined to
9
7͑d͒, respectively, with their respective vibrational term val-
ues of 1989, 1997, and 2042 cmϪ1, Table II. These term
values for overtones or combinations of ␯ and ␯ would be
be 1522 cmϪ1. A direct sum of the fundamental frequencies
of ␯ and ␯ is 1499 cmϪ1, and thus the anharmonic cou-
7
9
7
9
pling of ␯ and ␯ seems to be significant. However, the
7
9
FIG. 10. ͑a͒ Excitation spectrum of CH3NH2 for the origin band, ͑b͒ the
simulation with a stick spectrum, and ͑c͒ the simulation with an individual
linewidth of 14 cmϪ1. Parameters used in the simulation are listed in Table
III.
FIG. 8. Dependencies of molecular constants of excited CH3ND2 with the
number of ND2 wagging quanta: ͑a͒ ⌬G ϭG( ϩ1)ϪG( ), ͑b͒ B , ͑c͒
v
v
v
Ј
AT (ϭAF), and ͑d͒ V6 . Only reliable data are plotted here ͑see Table II͒.
Downloaded 01 Oct 2013 to 131.104.62.10. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jcp.aip.org/about/rights_and_permissions

J. Chem. Phys., Vol. 118, No. 24, 22 June 2003
Spectroscopy of methylamine. I
11035
TABLE III. Parameters used in the simulation for CH3NH2 excitation spectra.a
000
␯
␯
2␯
␯7ϩ␯
3␯
2␯
␯7ϩ2␯
9
9
7
9
9
9
7
AT
AF
9.57Ϯ0.10
4.93Ϯ0.10
͑0.78͒
14Ϯ1
0.38Ϯ0.03
͑5͒
8.54Ϯ0.1
4.86Ϯ0.1
͑0.77͒
13Ϯ1
0.41Ϯ0.03
͑19͒
͑9͒
͑6͒
͑0.77͒
12Ϯ1
0.44Ϯ0.03
͑20͒
͑8͒
͑4͒
͑0.76͒
16Ϯ2
͑10.4͒
͑4.8͒
͑0.77͒
12Ϯ2
͑7.2͒
͑4.8͒
͑0.74͒
18Ϯ3
0.29Ϯ0.06
͑7.2͒
͑4.8͒
͑0.74͒
18Ϯ3
0.29Ϯ0.06
͑10.4͒
͑4.8͒
͑0.77͒
Ͼ22
B
Ј
FWHM
␶ ͑ps͒
V6
0.33Ϯ0.04
0.44Ϯ0.08
Ͻ0.24
b
ZPE ͑torr͒
͑2.49͒
͑9.41͒
͑9.9͒
Evib
0
636Ϯ1
1008Ϯ1
1285Ϯ2
1549Ϯ10
1952Ϯ2
2014Ϯ5
2266Ϯ10
͓41669Ϯ1͔
aParameters in parentheses are less reliable.
bZero-point energy due to torsional motion.
quite helpful in the construction of potential energy surfaces
of this floppy molecule in its predissociative excited state
especially at near the top of the reaction barrier along the
specific bond-dissociation coordinate. Only accurately deter-
mined values for ⌬G ϭG( ϩ1)ϪG( ), B, AT, and V are
CH3NH2 in the symmetric-top approximation is assumed to
be 0.78 cmϪ1 compared to B ϭ0.67 cmϪ1 for CH3ND2 , de-
Ј
duced from the comparison of (BϩC)/2ϭ0.738 and 0.648
cmϪ1 of the electronically ground CH3NH2 and CH3ND2 ,
respectively.21 The torsional barrier height is presumed to be
identical, V6ϭ5 cmϪ1, to that of the CH3ND2 origin band.
Parameters used in the simulation with their uncertainties are
listed in Table III. The simulation with a stick spectrum rep-
resents the peak positions and relative intensities, Fig. 10͑b͒.
For fitting the real experimental spectrum, the Lorentzian
function with a FWHM of 14 cmϪ1 is convoluted to give the
simulation in Fig. 10͑c͒. The simulation matches the experi-
ment extremely well, and this indicates that the lifetime of
the excited CH3NH2 at the origin is ϳ0.38 ps. This perfect
match between experiment and simulation also validates the
Hamiltonian in Eqs. ͑1͒–͑4͒ employed in this work. The 0–0
transition frequency is accurately determined to be 41 669
cmϪ1 from the simulation.
v
v
v
6
plotted with the number of quanta in ND2 wag in Fig. 8. It is
interesting to note that both ⌬G and V show leveling offs
v
6
as the number of ND2 wag quanta increases, while B and AT
values are on a straight line. A detailed theoretical calcula-
tion of multidimensional potential energy surfaces would be
desirable to explain the experiment.
C. CH3NH2
The CH3NH2 excitation spectrum taken in the 41 500–
44 200 cmϪ1 range is shown in Fig. 9 with appropriate vi-
brational assignments. Similar to the case of CH3ND2 , ␯
7
(CH3 rock͒, ␯ (NH2 wag͒, and a combination of ␯ and ␯
9
7
9
bands are clearly identified in the CH3NH2 excitation spec-
trum. Spectral bandwidths, however, are much broader than
those of CH3ND2 , indicating much shorter lifetimes of
The first NH2 wagging band is very similar to the origin
band in its shape and structure, Fig. 11͑a͒. For the simulation
of this band, molecular constants determined for the ND2
wagging band of CH3ND2 are adjusted considering the
˜
CH3NH2 (A). The vibrational frequency of the amino-
NH2-to-ND2
substitution
effect. Accordingly,
AT
wagging mode is much blueshifted by D-to-H substitution of
the amino moiety, while the CH3 rocking frequency remains
more or less the same. The origin band of the CH3NH2 ex-
citation spectrum shown in Fig. 10͑a͒ is not rotationally well
resolved due to severe homogeneous line broadening. In the
first place, the CH3NH2 simulation can be tried by applying
the effect of the NH2-to-ND2 substitution to molecular con-
stants deduced from the CH3ND2 simulation. In CH3NH2 ,
AT and AF are quite distinct. Assuming that the minimum-
energy geometries of excited CH3ND2 and CH3NH2 are
identical, the moments of inertia of CH3 and NH2 are then
determined from those deduced from CH3ND2 to give
I (CH )ϭ3.42 amu Å2 and I (NH )ϭ1.71 amu Å2 ͑vide su-
ϭ8.54 cmϪ1 and AFϭ4.86 cmϪ1 are used in the simulation
Ј
Ј
3
2
pra͒, since the moment of inertia of NH2 is simply the
double of that of ND2 . Then ATϭ9.86 cmϪ1 and AF
ϭ4.93 cmϪ1 are expected. However, for the best fit of the
experiment, AT has to be decreased to 9.57 cmϪ1. This dif-
ference should come from different vibrational mode cou-
plings of CH3NH2 and CH3ND2 even at their origins ͑vide
infra͒. Accordingly, the internal rotational constants of AT
ϭ9.57 cmϪ1 and AFϭ4.93 cmϪ1 are used in the simulations
FIG. 11. ͑a͒ Excitation spectrum of CH3NH2 for the ␯ (NH2 wag͒ band,
9
͑b͒ the simulation with an individual linewidth of 13 cmϪ1, and ͑c͒ the
simulation with a stick spectrum. Parameters used in the simulation are
listed in Table III.
in Fig. 10. The overall rotational constant B of the excited
Ј
Downloaded 01 Oct 2013 to 131.104.62.10. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jcp.aip.org/about/rights_and_permissions

11036
J. Chem. Phys., Vol. 118, No. 24, 22 June 2003
Baek et al.
FIG. 12. ͑a͒ Excitation spectrum of CH3NH2 for the ␯ band, ͑b͒ the simu-
7
lation with an individual linewidth of 13 cmϪ1, and ͑c͒ the simulation with
a stick spectrum. Parameters used in the simulation are listed in Table III.
for the first NH2 wagging band. From the AT values deter-
mined at the origin and ␯ bands, AT and ␣T are calculated to
9
e
e
be 10.1 and 1.03 cmϪ1, respectively. It is very intriguing to
note that the ATe of CH3NH2 (10.1 cmϪ1) is nearly twice as
large as that of CH3ND2 (4.97 cmϪ1). This proves that
excited-state structures of CH3NH2 and CH3ND2 are almost
identical at their minimum potential energies. The overall
FIG. 14. ͑a͒ Lifetimes of excited CH3ND2 vs the vibrational energy, which
clearly show a strong mode dependence of the lifetime and ͑b͒ lifetimes of
excited CH3NH2 vs the vibrational energy. Note the different time scale on
vertical axes in ͑a͒ and ͑b͒.
rotational constant B is assumed to be 0.77 cmϪ1. A tor-
Ј
sional barrier height of 19 cmϪ1 is used, as in the CH3ND2
case. A linewidth of 13 cmϪ1 is used for the simulation in
Fig. 11͑b͒, and its stick spectrum is shown in Fig. 11͑c͒. The
agreement between experiment and simulation is excellent,
giving a lifetime of ϳ0.41 ps for the excited CH3NH2 at the
first NH2 wagging band. The fundamental frequency for the
NH2 wag is accurately determined to be 636 cmϪ1. This is
slightly larger than a corresponding value of 603.51 cmϪ1
from ab initio calculations.8
The ␯ band is very well isolated in the CH3NH2 exci-
7
tation spectrum, Fig. 12͑a͒. Highly excited internal rotor
states, compared to those of NH2 wag, are found to be less
intense. Accordingly, the molecular constants used in the
simulations in Figs. 12͑b͒ and 12͑c͒ are less reliable. How-
ever, the fundamental frequency of CH3 rock is accurately
determined in the simulation to give ␯7 (1–0)
ϭ1008 cmϪ1. Ab initio frequencies for ␯ and ␯ are re-
7
8
ported to be 1174 and 1051 cmϪ1, respectively.8 Thus simple
comparison of the experiment with ab initio calculation
could lead to a different assignment that the 1008 cmϪ1 band
is due to ␯8 . Previous studies of various isotope analogs of
methylamines provided solid evidence for the ␯ assignment
7
for this band.26 In addition, a frequency shift by NH2 /ND2
substitution of ␯ is expected to be much larger compared to
8
that of ␯7 ,8 and thus 1008 cmϪ1 is assigned as the CH3
rocking mode. This is very close to the value of 1012 cmϪ1
for the ␯ fundamental frequency of CH3ND2 . The fact that
7
the ␯ fundamental frequency of CH3ND2 is higher than that
7
of CH3NH2 could be due to different anharmonicites among
vibrational modes of two isotope species. Further spectro-
scopic works on CD3NH2 and CD3ND2 would be quite help-
ful in unraveling anharmonicities in great detail. The band-
width of 12 cmϪ1 is used in the ␯ band simulation in Fig.
7
12͑b͒ for the best match to the experiment, giving a lifetime
of ϳ0.44 ps, which is slightly longer than that of the origin
or first wagging band.
Experimental and simulated spectra for 2␯9 , ␯7ϩ␯9 ,
FIG. 13. Excitation spectrum ͑upper trace͒ with simulation ͑lower trace͒ of
3␯9 , 2␯7 , and ␯7ϩ2␯ bands are shown in Fig. 13. All
9
excited CH3NH2 for ͑a͒ 2␯9 , ͑b͒ ␯7ϩ␯9 , ͑c͒ 3␯9/2␯7 , and ͑d͒ ␯7ϩ2␯
9
peaks are found to be diffuse: thus, molecular constants re-
lated to overall and internal rotation are not expected to be
deduced accurately from simulations. However, associated
bands using parameters listed in Table II. In ͑c͒, the lower left trace is the
simulation for the 3␯ band, while the simulation for 2␯ is shown as the
9
7
lower right trace.
Downloaded 01 Oct 2013 to 131.104.62.10. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jcp.aip.org/about/rights_and_permissions

J. Chem. Phys., Vol. 118, No. 24, 22 June 2003
Spectroscopy of methylamine. I
11037
ciation of CH3ND2 occurs via tunneling through a barrier to
produce mainly DϩCH3ND(X) fragments. At the 2␯ en-
9
ergy level of CH3ND2 , the lifetime becomes much shorter,
giving ␶ϳ1.8 ps. This sudden decrease of the lifetime indi-
cates that predissociation of CH3ND2 may occur at near or
above the top of the barrier at a vibrational energy of ϳ1000
cmϪ1. Actually, the experimental fact that the lifetime is
ϳ0.38 ps at the origin of CH3NH2 and it varies little upon
increasing the vibrational energy, Fig. 14, strongly suggests
that predissociation of CH3NH2 may take place at near the
top of the barrier even at the zero-point energy level. The
difference in zero-point energies of CH3NH2 and CH3ND2
for the N–H͑D͒ stretching mode, which is the reaction coor-
dinate for the N–H͑D͒ bond cleavage, may account for the
difference of reaction barrier heights that CH3NH2 and
CH3ND2 experience along the N–H͑D͒ reaction coordinate.
This seems to be consistent with a conclusion by the Ashfold
group5,6 that the dominant predissociation channel of methy-
lamine is the N–H͑D͒ fission process. However, it should be
noted that our work confirms only that the major dissociation
FIG. 15. Simple diagram of potential energy surfaces along the N–H͑D͒
bond dissociation coordinate. Thermodynamically accessible reaction chan-
nels are shown on the right ͑Ref. 4͒. For the upper state, the adiabatic
potential energy surface is depicted in the vicinity of the Franck–Condon
region. Thereafter, it correlates to the HϩCH3NH(X) products diabatically
in the late stage along the reaction coordinate ͑Ref. 8͒.
channel is the N–H͑D͒ bond cleavage at the origin and ␯
9
vibrational energies are relatively quite accurately deter-
mined from simulations in Fig. 13, Table III. The vibrational
levels of the excited CH3ND2 and maybe at the origin level
of the excited CH3NH2 . At higher vibrational energies, how-
ever, the tunneling effect fades away as indicated in Fig. 14,
and thus other reaction channels are not conclusively ex-
cluded from our experimental results.
energy of 2␯ is determined to be 1285 cmϪ1. Using the
9
equation Evib( )ϭ␻ ( ϩ1/2)ϩ␹ ( ϩ 1)2 and a ␯ funda-
v
v
v
99
e
2
9
mental frequency of 636 cmϪ1, it is found out that
␻e (NH2)ϭ623 cmϪ1 and
␹
99 (NH2)ϭ6.5 cmϪ1
.
The
An interesting mode dependence of the lifetime is quite
clearly observed for CH3ND2 , Fig. 14͑a͒. That is, as clearly
␹
99 (NH2) is very similar to that of ND2 wag ͑vide supra͒.
These constants predict the vibrational energy of 3␯ to be
9
1947 cmϪ1, which is very close to the experimental value of
shown in Fig. 14͑a͒, lifetimes of ␯ and its combination
7
1952 cmϪ1, Table III. Vibrational energy term values for ␯
bands are about two times longer that those of isoenergetic
␯ overtone bands. The same trend is also observed in life-
9
7
ϩ␯9 , 2␯7 , and ␯7ϩ2␯ bands are found to be 1549, 2014,
9
and 2266 cmϪ1, respectively, while harmonic sums using ␯
times of CH3NH2 , though the mode dependence is less dra-
matic, Fig. 14͑b͒. A similar mode dependence on the excited
lifetime had been reported for N–H bond dissociation of
ammonia.9 For ammonia, thorough experimental and theoret-
ical studies were carried out to reveal fundamental aspects of
mode-dependent lifetimes.9–15 However, for methylamine,
our work provides mode-dependent lifetimes for the first
time. Both amino-wagging and methyl-rocking modes are
perpendicular to the reaction coordinates for bond dissocia-
tions. Thus, in the case of no intramolecular vibrational re-
distribution ͑IVR͒, these vibrational modes, which are per-
pendicular to the reaction coordinate, could survive until the
excited molecule reaches the transition state, giving vibra-
tional adiabatic potential energy surfaces characteristic to
each vibrational mode. In this case, the mode dependence of
the lifetime could be qualitatively explainable. The amino-
wag frequency is expected to be lowered at the transition
state compared to that of the excited reactant since one of the
N–H bonds gets lengthened at the transition state. Mean-
while, the CH3 rock frequency at the transition state is ex-
pected to be more or less same as that of the reactant, since
the CH3 group is not likely to be coupled to the N–H͑D͒
reaction coordinate. Therefore, in this case, the state excited
in the CH3 rocking mode will experience a higher effective
reaction barrier. For a better understanding of this interesting
mode dependence, multidimensional potential surfaces in the
vicinity of the transition state would be needed. More prac-
tically, calculation of two-dimensional potential energy sur-
7
and ␯ fundamentals give respective values of 1644, 2016,
9
and 2293 cmϪ1. These values for overtones or combinations
of ␯ and ␯ would be quite helpful in disentangling poten-
7
9
tial energy surfaces of this floppy molecule in its predisso-
ciative excited state. Lifetimes for these bands are also de-
duced from the simulations and listed in Table III.
D. Tunneling and predissociation
At the origin and ␯ bands, the excited CH3ND2 lifetime
9
͑ϳ8.8 ps͒ is about 20 times longer than the CH3NH2 lifetime
͑ϳ0.38 ps͒ at its origin, Fig. 14. This huge difference in
lifetimes of excited CH3NH2 and CH3ND2 is compelling
evidence of tunneling as a predissociation mechanism espe-
cially at the origin and first amino-wag bands. Other nonra-
diative processes such as internal conversion and intersystem
crossing do not seem to be responsible for the short lifetimes
of excited methylamines, since such electronic dephasing
processes are expected to be faster for CH3ND2 compared to
those for CH3NH2 due to the higher densities of low-lying
states of the former. Many reaction channels such as N–H,
C–H, and C–N bond dissociations are energetically acces-
˜
sible from methylamine in the A state, Fig. 15. However, the
huge effect on the predissociation rate by the H/D substitu-
tion of amino moiety is not expected in dissociation channels
not involving the N–H͑D͒ bond breakage. Therefore, at least
at the origin and first wagging vibrational levels, predisso-
Downloaded 01 Oct 2013 to 131.104.62.10. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jcp.aip.org/about/rights_and_permissions

11038
J. Chem. Phys., Vol. 118, No. 24, 22 June 2003
Baek et al.
are systematically measured here to give an interesting mode
dependence of the lifetime. Lifetimes of states with the
CH3-rocking mode excited are longer than those of isoener-
getic ND2-wagging modes. Multidimensional potential en-
ergy surfaces near the reaction barrier along the N–H͑D͒
reaction coordinate would be extremely helpful for a better
understanding of these interesting experimental facts in re-
gard to mode-dependent lifetimes. In conclusion, quantum
states of CH3NH2 and CH3ND2 in their first-excited states
are clearly characterized in this work. Since the excited mol-
ecule predissociates, methylamine serves as an excellent
model system for state-to-state reaction dynamics studies.
The effect of initial quantum reactant states on the reaction
outcome is an always-stimulating subject. Thus more dynam-
ics studies along this direction with theoretical calculations
are needed.
faces including amino wagging or CH3 rocking would be
desirable.
It is also interesting to note that, within the Born–
Oppenheimer approximation, the zero-point energy differ-
ence between CH3NH2 and CH3ND2 in the excited state
͑⌬ZPEЈ͒ is calculated to be 1018 cmϪ1 and this is much
smaller than that of 1387 cmϪ1 for the ground electronic
state ͑⌬ZPEЉ͒. The latter is from experimental values,21
while the former is calculated from the latter and origin en-
ergies of CH3NH2 and CH ND : ⌬ZPE ϭT (CH NH )
Ј
3
2
00
3
2
ϩ⌬ZPE ϪT (CH ND ) ϭ 41 669 ϩ1387Ϫ42 038 ϭ1018.
Љ
00
3
2
This huge difference between ⌬ZPEЈ and ⌬ZPEЉ is quite
rare, since the NH2 /ND2 substitution would affect the zero-
point energies almost equally for the ground and excited
states. Therefore, this indicates that potential energy surfaces
especially along N–H͑D͒ stretching coordinates are quite an-
harmonic in the excited state compared to those in the
ground state. That is, the more anharmonic excited potential
energy surface gives the lower zero-point N–H͑D͒ stretching
frequency, resulting in a decrease of the zero-point energy
difference between CH3NH2 and CH3ND2 . This anharmo-
nicity should come from severe coupling of the 3s–n Ryd-
berg state and repulsive valence state along the N–H͑D͒ dis-
sociation coordinate. Detailed ab initio calculations of the
coupling region of Rydberg and valence states would be de-
sirable for a quantitative explanation of this interesting ex-
perimental observation.
ACKNOWLEDGMENTS
This work was financially supported by Korea Research
Foundation Grant No. KRF-2002-070-C00046. The discus-
sion with Professor W. H. Miller at UC Berkeley was very
brief, but very helpful.
1 J. V. Michael and W. A. Noyes, J. Am. Chem. Soc. 85, 1228 ͑1963͒.
2 E. P. Gardner and J. R. McNesby, J. Phys. Chem. 86, 2646 ͑1982͒.
3 H. K. Haak and F. Stuhl, J. Phys. Chem. 88, 3627 ͑1984͒.
4 G. C. G. Waschewsky, D. C. Kitchen, P. W. Browning, and L. J. Butler, J.
Phys. Chem. 99, 2635 ͑1995͒.
5 C. L. Reed, M. Kono, and M. N. R. Ashfold, J. Chem. Soc., Faraday
Trans. 92, 4897 ͑1996͒.
IV. CONCLUSIONS
6 M. N. R. Ashfold, R. N. Dixon, M. Kono, D. H. Mordaunt, and C. L.
Reed, Philos. Trans. R. Soc. London, Ser. A 355, 1659 ͑1997͒.
7 T.-X. Xiang and W. A. Guillory, J. Chem. Phys. 85, 2019 ͑1986͒.
8 K. M. Dunn and K. Morokuma, J. Phys. Chem. 100, 123 ͑1996͒.
9 A. Bach, J. M. Hutchison, R. J. Holiday, and F. F. Crim, J. Chem. Phys.
116, 9315 ͑2002͒.
Here rovibrational structures and dynamics of methy-
lamines (CH3NH2 and CH3ND2) in predissociative A states
˜
(3s–n) are revealed through ͑1ϩ1͒ resonant-enhanced two-
photon ionization spectroscopy. Spectral origins are accu-
rately and precisely determined through the simulation to be
41 669 and 42 038 cmϪ1 for CH3NH2 and CH3ND2 , respec-
tively. The amino group which is bent in the ground elec-
tronic state becomes planar in the excited methylamine. Due
to this geometrical change upon electronic excitation, amino-
wagging and CH3-rocking modes are found to be optically
active, giving their respective fundamental frequencies of
636 ͑487͒ and 1008 ͑1012͒ cmϪ1 for CH3NH2 (CH3ND2). In
the excited state, CH3 rotates nearly freely about the C–N
10 D. H. Mordaunt, M. N. R. Ashfold, and R. N. Dixon, J. Chem. Phys. 109,
7659 ͑1998͒.
11 J. Biesner, L. Schnieder, J. Schmeer, G. Ahlers, X. Xie, K. H. Welge,
M. N. R. Ashfold, and R. N. Dixon, J. Chem. Phys. 88, 3607 ͑1988͒.
12 J. Biesner, L. Schnieder, G. Ahlers, X. Xie, K. H. Welge, M. N. R. Ash-
fold, and R. N. Dixon, J. Chem. Phys. 91, 2901 ͑1989͒.
13 L. D. Ziegler, J. Chem. Phys. 82, 664 ͑1985͒.
14 S. A. Henck, M. A. Mason, W.-B. Yan, K. K. Lehmann, and S. L. Coy, J.
Chem. Phys. 102, 4772 ͑1995͒; 102, 4783 ͑1995͒.
15 A. Nakajima, K. Fuke, K. Tsukamoto, Y. Yoshida, and K. Kaya, J. Phys.
Chem. 95, 571 ͑1991͒.
16 M. Kreglewski, J. Mol. Spectrosc. 72, 1 ͑1978͒.
˜
axis with respect to NH2(ND2). For CH3ND2 (A), torsional
17 N. Ohashi and J. T. Hougen, J. Mol. Spectrosc. 121, 474 ͑1987͒.
18 N. Ohashi, K. Takagi, J. T. Hougen, W. B. Olson, and W. J. Lafferty, J.
Mol. Spectrosc. 132, 242 ͑1988͒.
barriers of 5 and 19 cmϪ1 are accurately determined for the
zero-point and first amino-wagging levels, respectively,
through the simulation using the internal-rotor Hamiltonian
with a sixfold potential function. Lifetimes of quantum states
in excited methylamines are estimated from spectral line-
widths to give ϳ8.8 and ϳ0.38 ps for zero-point levels of
CH3ND2 and CH3NH2 , respectively. The huge NH2 /ND2
isotopic substitution effect on the excited-state lifetime
strongly indicates that the primary dissociation channel of
excited methylamine is the N–H͑D͒ bond dissociation and it
proceeds via tunneling through a reaction barrier at least at
low vibrational energies. At the vibrational energy of ϳ1000
cmϪ1, the lifetime of CH3ND2 suddenly drops to 1.8 ps,
suggesting a comparable reaction barrier height along the
N–D reaction coordinate. Lifetimes of excited methylamines
19 D. Kivelson, J. Chem. Phys. 27, 353 ͑1957͒.
20 M. Kreglewski, J. Mol. Spectrosc. 133, 10 ͑1989͒.
21 A. P. Gray and R. C. Lord, J. Chem. Phys. 26, 690 ͑1957͒.
22 M. Oda, N. Ohashi, and J. T. Hougen, J. Mol. Spectrosc. 142, 57 ͑1990͒,
and references therein.
23 K. Shimoda, T. Nishikawa, and T. Itoh, J. Chem. Phys. 22, 1456 ͑1954͒.
24
¨
T. Forster and J. C. Jurgens, Z. Phys. Chem. Abt. B 36, 387 ͑1937͒.
25 E. Tannenbaum, E. M. Coffin, and A. J. Harrison, J. Chem. Phys. 21, 311
͑1953͒.
26 M. Tsuboi, A. Y. Hirakawa, and H. Kawashima, J. Mol. Spectrosc. 29, 216
͑1969͒.
27 M. Tsuboi and A. Y. Yirakawa, Can. J. Phys. 60, 844 ͑1982͒.
28 D. P. Taylor and E. R. Bernstein, J. Chem. Phys. 103, 10453 ͑1995͒.
29 D. P. Taylor, C. F. Dion, and E. R. Bernstein, J. Chem. Phys. 106, 3512
͑1997͒.
30 M.-J. Hubin-Franskin, J. Delwiche, A. Giuliani, M.-P. Ska, F. Motte-
Downloaded 01 Oct 2013 to 131.104.62.10. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jcp.aip.org/about/rights_and_permissions

J. Chem. Phys., Vol. 118, No. 24, 22 June 2003
Spectroscopy of methylamine. I
11039
Tollet, I. C. Walker, N. J. Mason, J. M. Gingell, and N. C. Jones, J. Chem.
Phys. 116, 9261 ͑2002͒.
33 R. H. Judge and D. J. Clouthier, Comput. Phys. Commun. 135, 293
͑2001͒.
34 S. J. Baek, K.-W. Choi, Y. S. Choi, and S. K. Kim, J. Chem. Phys. 118,
11040 ͑2003͒, following paper.
31 S. J. Baek, K.-W. Choi, Y. S. Choi, and S. K. Kim, J. Chem. Phys. 117,
10057 ͑2002͒.
35 T. J. Sears, P. M. Johnson, and J. BB-Wang, J. Chem. Phys. 111, 9213
͑1999͒.
32 P. R. Bunker and P. Jensen, Molecular Symmetry and Spectroscopy ͑NRC
Research, Ottawa, 1998͒.
36 P. M. Johnson and T. J. Sears, J. Chem. Phys. 111, 9222 ͑1999͒.
Downloaded 01 Oct 2013 to 131.104.62.10. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jcp.aip.org/about/rights_and_permissions