HCN2+ dication spectroscopy: Theoretical and experimental investigations

Article abstract of DOI:10.1016/j.chemphys.2004.09.026

The present work combines experimental and theoretical investigations of HCN²⁺ to investigate the spectroscopy and dissociation dynamics of this dication. Its spectrum is a complex blend of overlapping features, consistent with theoretical calculations showing that the low-lying electronic states of HCN²⁺ exhibit Renner-Teller and vibronic interactions. The double ionisation spectra are dominated by C-N stretching excitations for all the electronic states investigated. Long-lived HCN²⁺ ions are observed and are believed to be in low vibrational levels of the X ³Σ^- and/or ¹Δ and/or ¹Σ⁺ electronic states; some have metastable lifetimes of about 25 ns. The formation of H⁺ + CN⁺ products is attributed to predissociation by the lowest ³A″ component resulting from a conical intersection between the X ³Σ^- and ³Π states. The overall predissociation processes involve vibronic, Renner-Teller and spin-orbit couplings between several electronic states of HCN²⁺.

Full text of DOI:10.1016/j.chemphys.2004.09.026

Chemical Physics 309 (2005) 291–301
www.elsevier.com/locate/chemphys
2
+
HCN dication spectroscopy: theoretical and
experimental investigations
a,
b
b
*
M. Hochlaf , A. Pilcher-Clayton , J.H.D. Eland
a
Theoretical Chemistry Group, University of Marne-la-Vall e´ e, Champs sur Marne, F-77454 Marne-la-Vall e´ e Cedex 2, France
b
Physical and Theoretical Chemistry Laboratory, South Parks Road, Oxford OX1 3QZ, UK
Received 30 July 2004; accepted 29 September 2004
Available online 22 October 2004
Abstract
2
+
The present work combines experimental and theoretical investigations of HCN to investigate the spectroscopy and dissocia-
tion dynamics of this dication. Its spectrum is a complex blend of overlapping features, consistent with theoretical calculations
2
+
showing that the low-lying electronic states of HCN exhibit Renner–Teller and vibronic interactions. The double ionisation spec-
2
+
tra are dominated by C–N stretching excitations for all the electronic states investigated. Long-lived HCN ions are observed and
3
ꢀ
1
1 +
are believed to be in low vibrational levels of the X R and/or D and/or R electronic states; some have metastable lifetimes of
+
+
3
00
about 25 ns. The formation of H + CN products is attributed to predissociation by the lowest A component resulting from a
conical intersection between the X R and P states. The overall predissociation processes involve vibronic, Renner–Teller and
spin–orbit couplings between several electronic states of HCN
2004 Elsevier B.V. All rights reserved.
3
ꢀ
3
2
+
.
ꢀ
1
. Introduction
tra [6]. Some aspects of the dissociation of HCN
dications have been investigated by the photoelectron
photoion–photoion coincidence (PEPIPICO) technique
[7], but without energy selection of the initial states.
The hydrogen cyanide molecule is an important con-
stituent of interstellar clouds [1] and planetary atmos-
pheres [2–4], and is possibly a precursor of life on
Earth [5]. Its significance in many environments makes
knowledge of all its spectroscopic properties very desir-
able. The neutral molecule and singly charged positive
ion have already been studied in depth; here we present
a first detailed investigation of the doubly positive ion
HCN , using the time-of-flight photoelectron–photo-
electron coincidence technique (TOF-PEPECO) and
high level calculations.
2
þ
2þ
The isoelectronic species N and C H both show
2
2
2
well-resolved spectra in experiments by the TPEsCO
and PEPECO techniques [8–10], so we could hope for
a similar outcome from work on HCN. However, expe-
+
rience of HCN monocation spectroscopy sounds a
warning note.
2
+
The photoelectron spectrum of HCN, although meas-
ured with good resolution at an early date [11], is noto-
rious for the difficulty of its interpretation. Two
2
+
2
2 +
The existence of HCN dications with lifetimes of at
least microseconds has been known for some time be-
cause of the presence of peaks at m/q 13.5 in mass spec-
electronic states, X P and A R are close in energy
(separation 0.4 eV) and interact vibronically, producing
a spectrum which cannot be analysed by pattern recog-
nition methods into systems of independent progres-
sions. The spectrum of the two lowest states was
explained only after detailed calculations incorporating
the vibronic coupling [12] and its modulation by other-
wise inactive modes [13]. The second excited stateÕs spec-
*
Corresponding author. Tel.: +33 1 60 95 73 19; fax: +33 1 60 95
7
3 20.
E-mail addresses: hochlaf@univ-mlv.fr (M. Hochlaf), eland@
physchem.ox.ac.uk (J.H.D. Eland).
0
301-0104/$ - see front matter ꢀ 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.chemphys.2004.09.026

292
M. Hochlaf et al. / Chemical Physics 309 (2005) 291–301
trum has been elucidated more recently [14], and proved
to be almost equally complex because of interaction be-
tween P and R components. Simple considerations of
rectly proportional to the ionsÕ momenta. The condi-
tions of the measurements favour detection of low
energy electrons, so the relative intensity of doubly ion-
ised products is exaggerated in the data.
HCN was prepared by heating an intimate mixture of
carefully dried NaCN and stearic acid to just above the
2
+
the orbital occupancy predict that in HCN , the two
2
2
configurations involving the outermost orbitals, 5r 1p
1
3
and 5r 1p will give rise to six electronic states within
a few electronvolts. The spectrum may, therefore, be
no less complex than that of HCN and could be more
melting point, 70 ꢁC, in vacuum. Impurity CO was re-
2
+
moved by pumping on the condensed product cooled in
complicated.
a CO /acetone slush bath. DCN was prepared by drop-
2
As shown below, the experimental spectra are struc-
tured, but very difficult to interpret. Electronic structure
calculations at the highest level are therefore essential
for interpretation of the experimental spectra, and even
with their input, uncertainty remains. From considera-
ble recent experience in calculations of molecular dica-
tion spectra [8,15–18] and successful comparison with
experiment, we can perhaps have more confidence in
the theoretical results per se than in the fit to the exper-
imental data.
wise addition of D O to a mixture of NaCN and P O
5
under nitrogen with collection in a cold trap. It was
purified in the same way. The whole apparatus was con-
2
2
ditioned with D O before the DCN runs to avoid forma-
2
tion of HCN by isotopic exchange in the gas inlet
system. The purity of the DCN was checked by measur-
ing its PES spectrum at 21 eV photon energy; the spec-
trum was consistent with that of the pure compound
[11].
2
. Experimental method
3. Theoretical methods
The basis of the TOF-PEPECO method [19] is detec-
Electronic correlations are expected to play an impor-
tant role in the formation and in the metastability of
molecular dications. In previous work dealing with tri-
atomic dications, we have shown that highly correlated
ab initio methods are needed in the calculations of the
potential energy functions (PEFs) and of the properties
tion and energy analysis of both electrons of each pair
ejected in double photoionisation. The light source is a
pulsed discharge lamp with helium or neon as the work-
ing gas; in its present form it gives ca. 5 ns pulses at 9
kHz repetition rate. Ionisation occurs where wave-
length-selected light is focussed into an effusive gas jet
at the focus of a long (5.5 m) magnetic bottle time-of-
flight spectrometer. Further details have been given in
recent papers [9,20].
3
ꢀ
of such species. [16–18]. Here, the PEFs of the X R ,
3
1
1
+
1
2+
P, D, R and P states of HCN have been mapped
using such highly correlated approaches, for instance the
complete active space self-consistent field (CASSCF)
[21] and the internally contracted multi-reference config-
uration interaction (MRCI) methods [22,23]. For these
calculations, the generally contracted spdf (spd) cc-
pV5Z basis set of Dunning has been used for C/N (H)
atoms, resulting in 156 contracted Gaussian functions
[24]. In order to reduce symmetry breaking problems,
the CASSCF computations started with the HF molec-
Double ionisation spectra were accumulated at two
wavelengths, 30.4 and 25.6 nm, corresponding to the
photon energies 40.8 and 48.4 eV. Calibration was by
comparison with photoelectron spectra of O2 taken
immediately before and after each run. Because the
detection times of all electrons are recorded in the list
mode, contact potential changes during long run times
could be compensated in the data reduction. Spectra
from each few hours of runtime were cross-correlated
and optimum shifts determined before addition to form
the final spectra. The experimental resolution is a func-
tion of electron energy, and also depends on the accu-
racy of the calibration. The narrowest lines in the
1
+
ular orbitals obtained for HCN(X R ). In the active
space, all valence molecular orbitals were optimized.
In these calculations, electronic states having the same
spin multiplicity were averaged together. All valence
electrons were correlated. For MRCI calculations, all
configurations in the CI expansion of the CASSCF
wavefunctions were taken into account as a reference.
2
+
spectrum of HCN at 40.8 eV photon energy are about
5
0-meV wide.
The calculations were done in C symmetry with wave-
s
6
PEPIPICO spectra (without electron energy analysis)
functions corresponding to more than 31 · 10
6
were measured at the same wavelengths in connection
with an earlier investigation [7], where the apparatus
was described. In essence, ions and ion pairs of all
masses were analysed by time of flight, in coincidence
with photoelectrons ejected from the molecules by mon-
ochromatic light. Under the focus conditions used, the
widths of peaks in the ion time-of-flight spectra are di-
(54 · 10 ) uncontracted configurations for the singlets
(for the triplets). All electronic calculations were per-
formed using the MOLPRO program suite [25].
2
+
3
ꢀ
The three-dimensional PEFs of the HCN (X R ,
3
1
1
+
1
P, D, R and P) have been mapped in the internal
coordinates of this dication corresponding to the CN
(RCN), CH (RCH) stretches and to the in-plane bending

M. Hochlaf et al. / Chemical Physics 309 (2005) 291–301
293
2
+
angle (h). For the singlets, the calculations were per-
tify the main band with states of HCN
from the
2
2
2
2
1
3
formed for 22 different geometries around the equilib-
rium minimum of each electronic state covering the
configurations 4r 5r 1p and 4r 5r 1p , and the sub-
1
2
3
1
1
4
sidiary band with states from 4r 5r 1p or 4r 5r 1p .
Resolved structure is visible only at the lower photon
energy in Fig. 1 because in the TOF-PEPECO method
the energy resolution degrades at higher electron ener-
gies. Details of the partially resolved structures are
shown for both the HCN and DCN molecules in Fig.
2, and the energies of the distinguishable lines are tabu-
lated in Table 1. Although relative energies of the lines
are more precise (± 0.005 eV), the absolute accuracy of
the calibrations is not better than ± 0.03 eV. This means
that isotope shifts in ionisation potential between HCN
and DCN cannot be used to identify vibrational origins.
ꢀ
1
energy ranges up to ꢁ5000 cm above each minimum.
For the triplets, the PEFs have been carried out for 50
ꢀ1
different geometries up to energies ꢁ15,000 cm above
2
+
3
ꢀ
the HCN (X R ) minimum, so that the equilibrium
and the conical intersection regions (see below) are both
covered. The calculated energies were fitted to polyno-
mial expansions in displacement coordinates for both
the stretches and the bending. The resulting PEF expan-
sions were used to calculate the quartic force fields in
internal coordinates, which have been transformed by
the l-tensor algebra to quartic force fields in dimension-
less normal coordinates [26,27]. These data allow us to
evaluate a set of spectroscopic properties using second-
order perturbation theory. The PEF expansions were
also used in variational calculations using the approach
of Carter and Handy [28].
2
+
2+
The spectra of both HCN and DCN are systems
of congested vibrational bands, overlying a continuum
whose relative intensity increases towards higher energy.
There is no unambiguous onset, but a gradual rise in sig-
nal from about 37 eV leads to the first clear band(s) near
37.5 eV. At higher energies, sequences of peaks in the
spectra of both isotopomers form progressions, from
which average spacings can be extracted. In the range
38.5–39.2 eV, there seem to be single progressions with
As part of this work, the three-dimensional PEF of
1
+
the ground state X R of HCN has been generated at
the CCSD(T) level of theory and using the same basis
set as for the dication calculations. These calculations
are not presented here, but the calculated fundamentals
ꢀ
1
mean intervals of 1090 ± 80 cm for both HCN and
DCN. The spectra of the isotopomers differ more in
the lower energy range, where there are signs of inter-
penetrating progressions or more complex structures.
In addition to spectra of dication states, the TOF-
PEPECO method gives complete energy distributions
of the electrons emitted in double photoionisation.
Fig. 3 shows a two-dimensional map for HCN at 40.8
eV photon energy. Distributions of electrons are shown
in the left part of the figure for formation of selected
HCN levels. For some of the higher levels, there is a
broad peak at 0.15 eV, which presumably represents a
contribution from autoionisation of an intermediate
state of HCN near the photon energy. In formation
of the HCN levels below 37.5 eV, the distribution
shows only a gentle slope favouring lower electron ener-
ꢀ1
differ by less than 20 cm from the experimental values
29]. This ground-state PEF can be provided on request.
[
It has been used here to evaluate Franck–Condon fac-
tors for the direct double photoionisation process.
4
. Results
4
.1. Results from TOF-PEPECO experiments
2
+
Fig. 1 shows the double photoelectron spectra of
HCN at the two photon energies, 40.8 and 48.4 eV.
There is clearly a main group of states between 37 and
+
2
+
4
1 eV, followed by a weaker feature near 45 eV. By com-
parison with the single ionisation spectrum, we can iden-
Fig. 1. Overall double photoionisation spectra of HCN at 40.8 and
Fig. 2. Detail of the structured parts of the double photoionisation
spectra of HCN and DCN at 40.8 eV photon energy.
48.4 eV photon energy.

294
M. Hochlaf et al. / Chemical Physics 309 (2005) 291–301
Table 1
at each photon energy gives the fraction of double ion-
isation. This is found to be just 0.2% at 40.8 eV and rises
to 5.1% at 48.4 eV photon energy.
Energies of distinguishable peaks in the TOF-PEPECO spectra. Where
three decimal places are quoted relative precision is ± 0.005 eV, but
absolute accuracy is only ± 0.03 eV
Line
HCN
DCN
4
.2. Results from PEPIPICO experiments
1
1
2
3
3
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
2
a
b
37.486
37.546
37.710
37.862
37.936
38.01
38.104
38.289
38.464
38.572
38.697
38.847
38.974
39.105
39.246
39.36
37.517
37.566
37.738
37.889
37.948
Experiments on HCN and DCN by the PEPIPICO
technique were carried out some time ago in connection
with a different investigation [7], but only limited results
were published at that time. Measurements on both
HCN and DCN were made at the same two photon
energies used in the current TOF-PEPECO work, so
the results refer to exactly the same photoionisation
processes. Fig. 4 shows the PEPIPICO signatures of
the principal dication dissociation reaction, which pro-
duces H + CN in both compounds at 40.8 eV. The
diagonal ‘‘tails’’ are due to the decay of metastable dica-
tions, whose existence in HCN has already been re-
marked upon [31].
From comparison between the simultaneously meas-
ured mass spectra and PEPIPICO spectra with the
known ion detection efficiency, we can estimate the pro-
portions of different double ionisation channels. Within
the double ionisation, the fraction observed as parent
a
b
c
38.127
38.310
38.47
38.588
38.724
38.858
38.998
39.129
39.29
0
1
2
3
4
5
7
8
9
0
+
+
2
+
39.49
39.61
39.73
39.84
39.96
40.08
2
+
HCN ions (rather than ion pairs) is 8 ± 2% at both
photon energies. The ion pairs produced are almost all
gies; most other apparent structures are statistical noise
from the low numbers of counts. There is no very clear
sign of the characteristic electron energies (0.22, 0.34
and 1.01 eV) from autoionisation of atomic N* as seen
+
+
+
+
H + CN (95%) at 40.8 eV, with just 5% CH + N .
At 48.4 eV more different pairs are formed, but the sim-
ple H–C cleavage still provides the majority (77%).
From the widths of the ion pair coincidence peaks,
the kinetic energy release and aspects of its distribution
can be determined [32]; because of kinematic factors, the
precision is best in DCN as the peak is wider. At both
40.8 and 48.4 eV photon energy, the total energy release
in other compounds with terminal N atoms (N and
2
N O) [9,30], though the first peak near 0.21 eV may pos-
2
sibly be present.
Finally, because the electron collection efficiency is
known, the observed ratio of pairs to single electrons
Fig. 3. Complete photoelectron energy distribution map (right panel) for HCN double photoionisation at 40.8 eV, with projections of the
distributions along lines of fixed electron pair energy, corresponding to selected final state energies. The distributions are essentially flat, indicating
that direct photoionisation is the major pathway.

M. Hochlaf et al. / Chemical Physics 309 (2005) 291–301
295
+
+
+
+
Fig. 4. PEPIPICO distributions for the charge separation reactions, at 40.8 eV photon energy, forming H + CN (right) and D + CN (left). The
diagonal tails demonstrate the existence of metastable dications with lifetimes of the order of 20 ns (see text).
is sharply peaked at 5.4 ± 0.2 eV. If the products are
tion energies are listed in Table 2, where the accuracy is
expected to be better than 0.1 eV for such large ab initio
calculations. In the 0–8 eV internal energy range, several
bound states in both coordinates are found. These elec-
tronic states present several avoided crossings such as
formed in their electronic and vibrational ground states
+
(
or in the almost degenerate first excited state of CN ),
2
+
they must originate in levels of HCN
around
3
3.1 + 5.4 = 38.5 eV, where 33.1 eV is the lowest disso-
+
+
1
+
1 +
ciation limit for H + CN derived from the standard
heats of formation. This result refers to rapid (<10 ns)
dissociations into H + CN ; the energy release in the
slower metastable decays cannot be determined from
the data.
the one between the 1 R and 2 R states at R ꢁ4
CH
bohr responsible for the potential barrier in the lower
one (cf. Fig. 5(a)). One can mention also the avoided
+
+
3
crossings between the P states which give rise to the lo-
cal minima of these electronic states along both RCH and
RCN coordinates. Additionally, we notice the existence
of conical intersections between the electronic states of
the dication. The most obvious one is the conical inter-
From the form of the metastable ‘‘tails’’, it is possible
to determine the metastable ionsÕ lifetime [31]. From
simple inspection of the arrays, we see a clear tail (Fig.
3
ꢀ
3
4
), but there is no ‘‘V’’ in the notation of [31]. This indi-
section between the X R and the 1 P states along the
CN stretch at RCN ꢁ2.3 bohr (see Fig. 6(b)). As already
mentioned in Section 1, such vibronic interactions are
expected, since the lowest molecular orbitals of HCN
are close in energy. Nevertheless, the intersection be-
cates that there is no long (microsecond) lifetime compo-
nent. The form of the tails is quite close to exponential,
giving mean lifetimes (after coordinate transformation)
2
+
2+
of 34 ± 5 ns for DCN and 20 ± 8 ns for HCN at
0.8 nm. Although single mean lifetimes fit the tail data
2
+
4
tween the two electronic states of HCN is located at
2
+
3
ꢀ
well, it is still probable that multiple lifetimes are actu-
ally present.
ꢁ0.7 eV above the HCN (X R ) minimum, which is
high enough to allow us to calculate the vibrational
2
+
structure of the electronic ground state of HCN with-
out considering this vibronic interaction explicitly (cf.
Section 4.3.2). Several other conical intersections can
4
4
.3. Results from ab initio calculations
2
+
.3.1. Excited states
Fig. 5(a) and (b) gives an overview of MRCI collinear
be found between the excited states of HCN , for in-
1
ꢀ
1
stance the one between 1 R and 2 P at R ꢁ3 bohr,
CH
3 + 3
potential energy curves for the electronic excited states
1 R and 2 P at R ꢁ4 bohr. Furthermore, all the
CH
2
+
of HCN
RCH coordinate, where the CN bond length is fixed at
.53 bohr, corresponding to its equilibrium value in
of singlet (triplet) multiplicity along the
doubly degenerate states (i.e., all except the R states)
are expected to split into two components for bent struc-
tures because of the Renner–Teller effect. The vibronic
and Renner–Teller couplings mix these electronic states
together, complicating the mapping of their three-
dimensional potential energy functions and the calcula-
tions of their rovibronic spectra. According to the curves
of Figs. 5 and 6, the experimental spectra are expected to
2
2
+
3
ꢀ
HCN (X R ). Fig. 6 presents the same states as Fig.
for the collinear configuration where RCN is varied
and RCH is kept fixed at 2.205 bohr (ꢁits equilibrium
5
2
+
3
ꢀ
value in HCN (X R )). These PEFs are given relative
2
+
3
ꢀ
to the minimum of HCN (X R ). The vertical excita-

296
M. Hochlaf et al. / Chemical Physics 309 (2005) 291–301
1
2
∆
1
Π
3
3
1 ∆
8
6
4
2
0
8
6
4
2
0
1
+
3
Σ
1
-
3
+
2
Σ
2 Σ
1
2
Π
3
3
Π
1
-
1 Σ
3
+
1
Σ
3
2
Π
1
+
2
Σ
1
Π
1
1
+
1
Σ
3
1
1 Π
1
∆
3
-
X Σ
RCH 2.53
++
RCH 2.53
++
N]
[H
C
N]
[H
C
2
3
4
5
6
2
3
4
5
6
(
a)
RCH / bohr
(b)
RCH / bohr
2+
Fig. 5. MRCI collinear potential energy curves of the singlet (a) and triplet (b) electronic states of HCN along the RCH coordinate. The RCN
distance is set to 2.53 bohr.
1
+
3
+
3
Σ
2 Σ
3
+
8
6
4
2
0
8
6
4
2
0
2 Σ
3
2
∆
1
2
Π
3
3
3
Π
1
∆
1
+
2
Σ
1
2 ∆
3
2
Π
3
+
3
-
1
Σ
2
Σ
1
-
1 Σ
1
1
Π
3
Π
1
1
+
1
Σ
1
1
∆
3
-
X Σ
2
.205 RCN
2.205 RCN
++
N]
++
N]
[H
C
[H
C
2
3
4
5
6
2
3
4
5
6
RCN / bohr
RCN / bohr
(
a)
(b)
2+
Fig. 6. MRCI collinear potential energy curves of the singlet (a) and triplet (b) electronic states of HCN along the RCN coordinate. The RCH
distance is set to 2.205 bohr.

M. Hochlaf et al. / Chemical Physics 309 (2005) 291–301
297
Table 2
Equilibrium geometries, MRCI vertical excitation energies (T), zero point vibrational energies (G
terms (a )) of the excited states of HCN
i
0
e
) and rotational constants (B ; vibration–rotation
2+
˚
˚
a
ꢀ1
ꢀ1
ꢀ1
ꢀ1
ꢀ1
State
R
CH (A)
R
CN (A)
T (eV)
G
0
(cm
)
B
e
(cm
)
a
1
(cm
)
a
2
(cm
)
a
3
(cm
)
3
ꢀ
b
X R
1.16
1.16
1.16
1.15
1.21
1.34
1.23
1.36
1.36
1.24
0
2381
–
2450
2479
2383
1.1093
0.0113
ꢀ0.0055
0.0135
3
d
d
d
c
P
0.67
1
D
R
0.87
1.28
2.52
1.0920
1.0893
1.2526
1
1
+
0.0109
ꢀ0.0027
0.0127
P
a
b
c
2+
3
ꢀ
It corresponds to the energy difference between the minimum of the HCN (X R ) state and the minimum of the upper state.
2+
3
ꢀ
Total MRCI energy at the equilibrium geometry of HCN (X R ): ꢀ91.94463621 a.u.
2+
Perturbations due to the vibronic interaction between this electronic state and the ground state of HCN (see text).
d
3
0
From the A component.
2
+
3
ꢀ
be very complicated for internal energies > 4 eV. Finally,
we remark that the crossings and avoided crossings be-
tween the electronic states of HCN , together with
state slightly. In order to deduce the HCN (X R )
spectroscopic properties, we have mapped the 3D-PEF
2
+
2+
3
ꢀ
of HCN (X R ) close to its equilibrium geometry,
1
1
the dissociative character of some of them (e.g., the
well below this conical intersection. For the D and P
3
1
0
2
P state) will probably play a crucial role on the meta-
states, the quartic force fields of both their A and
A components are given in Table 3. However, only
the rovibrational spectrum using one component
(the A one) is presented here without considering the
Renner–Teller interaction since the main aim of these
calculations is to propose an interpretation for the fea-
tures observed in the TOF-PEPECO spectra. The differ-
ences between the energies of the rovibrational levels
given here and those calculated using full treatment of
the Renner–Teller and vibronic couplings are expected
1
00
stability of this dication.
1
00
4
.3.2. Spectroscopy
Table 3 lists the quartic force fields of HCN
2
+
3
ꢀ
1
1 + 1
(
X R , D, R , P) in dimensionless normal coordi-
nates. All these electronic states are found to be linear.
3
ꢀ
3
The conical intersection between the X R and the P
states, mentioned above, is expected to complicate
the pattern of rovibrational levels in its vicinity
3
ꢀ
3
ꢀ1
(
i.e., the upper levels of X R and of P) and should
to be less than 30 cm . [33] The accuracy of the present
affect the low-lying rovibrational levels of the ground
theoretical results is comparable with that of the exper-
imental spectra of these and similar polyatomic dica-
tions. For example, readers are referred to other recent
comparisons between theoretical and experimental
results on other triatomic dications [16–18].
Table 3
2+
3
ꢀ
1
1 + 1
Quartic force fields of HCN (X R , D, R , P) in dimensionless
normal coordinates (all values are in cm ). See text for more details
ꢀ1
From the values given in Table 3, it is evident that
there is coupling between the CH stretch and the bend-
ing modes for all the investigated electronic states, since
both the third and fourth-order force-field terms relative
to these modes (for instance /122 and /1122) have large
values. Only weak coupling is expected between the
3
ꢀa
1
b
1
+
1
b
X R
D
R
P
x
x
x
1
2
3
2498.9
516.7
1307.6
ꢀ570.3
ꢀ1906.7
170.1
2543.5
579.2 (615.5)
1290.5
2557.8
608.7
1250.6
ꢀ527.4
ꢀ2063.0
84.3
2121.4
575.2 (713.7)
1483.7
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
333
ꢀ572.4
ꢀ898.5
111
ꢀ2051.7
928.2
1
two stretching modes except for the P state, where
223
120.5 (108.2)
907.7 (843.6)
94.8
53.5 (7.2)
ꢀ317.9 (ꢀ392.7)
474.0
the /133, /113 and /1113 terms are quite large.
122
993.0
97.0
872.8
83.2
133
Tables 4–6 list the rovibrational theoretical spectra of
2
+
3
ꢀ
1
1 + 1
113
ꢀ277.7
1345.7
168.7
1120.0
ꢀ59.2
ꢀ35.0
ꢀ117.1
ꢀ1090.1
25.9
ꢀ266.3
ꢀ238.9
1048.8
134.7
1073.7
ꢀ32.8
ꢀ23.3
ꢀ87.5
ꢀ1001.2
14.1
ꢀ1155.3
HCN (X R , D, R , P). The calculations are car-
ried out for J = 0 and 1 for the R electronic states and
only for J = 0 for the D and P states. For these elec-
tronic states, anharmonic resonances can be found be-
tween their rovibrational levels even for the ones
2222
3333
1111
2233
1333
1223
1122
1169.3 (922.4)
154.2
35.5 (193.8)
402.5
1085.3
787.6
ꢀ52.1 (ꢀ44.0)
ꢀ258.7 (ꢀ226.7)
ꢀ31.2
ꢀ368.6
ꢀ
1
ꢀ100.1 (ꢀ94.6)
ꢀ1045.2 (ꢀ979.5)
20.2
423.2 (353.9)
ꢀ608.9(ꢀ456.3)
537.7
located as low as 2000 cm . For example, the two levels
2+ 1
+
ꢀ1
of HCN
R located at 1186 and 1240 cm are mix-
1133
1113
a
ture of (0,0,1) and (0,2,0) vibrational wavefunctions with
equal weights. Such resonances mix the rovibrational
wavefunctions of these levels making their assignment
by quantum numbers quite difficult. For these reasons
the attributions given in Tables 4–6 are only tentative.
142.8
125.1
115.1
ꢀ582.5
This PEF is mapped well below the conical intersection, so that
this quartic force field is valid only close to the equilibrium geometry of
2+
3
ꢀ
HCN (X R ).
b
1
00
Between parentheses are given the values for the A component.

298
M. Hochlaf et al. / Chemical Physics 309 (2005) 291–301
Table 4
Rovibrational levels of HCN (X R ). The Franck–Condon factors
Table 6
2+
3
ꢀ
2+ 1
+
ꢀ1
Rovibrational levels of HCN ( R ) (in cm ). The Franck–Condon
factors (FCF) corresponding to the double ionisation from
HCN(X R ) are also given
1
+
(FCF) corresponding to the double ionisation from HCN(X R ) are
1
+
also given
+
+
r
(J = 0)
,v
p (J = 1)
(v ,v ,v
r
(J = 0)
,v ,v
p (J = 1)
(v ,v ,v
ꢀ1
ꢀ1
ꢀ1
ꢀ1
Energy (cm )
(
v
1
,v
2
3
)
Energy (cm
)
FCF
1
2
3
)
Energy (cm
)
(v
1
2
3
)
Energy (cm
)
FCF
1
2
3
)
(
(
(
(
(
(
(
(
(
(
(
(
(
(
.
0,0,0)
0,2,0)
0,0,1)
0,4,0)
0,2,1)
1,0,0)
0,0,2)
0,6,0)
0,4,1)
1,2,0)
0,2,2)
1,0,1)
0,0,3)
0,8,0)
. .
0
1009
1278
2008
2293
2348
2531
2998
3294
3328
3552
3612
3766
3981
0.0256
(0,1,0)
(0,3,0)
(0,1,1)
(0,5,0)
(0,3,1)
(1,1,0)
(0,1,2)
(0,7,0)
(0,5,1)
(1,3,0)
510
1514
1792
2508
2800
2842
3049
3495
3792
3822
(0,0,0)
(0,2,0)
(0,0,1)
(1,0,0)
(0,2,1)
(0,0,2)
(0,4,0)
(1,2,0)
(0,2,2)
(0,0,3)
(1,0,1)
(0,2,2)
(0,6,0)
(2,0,0)
(1,4,0)
(1,2,1)
(0,0,4)
(1,0,2)
(0,2,3)
0
1186
1240
2341
2358
2417
2490
3493
3516
3546
3589
3652
3754
4592
4627
4665
4682
4748
4763
4812
4902
5029
5725
5742
5781
5800
5812
5883
5891
0.017
(0,1,0)
(0,3,0)
(0,1,1)
(1,1,0)
(0,3,1)
(0,1,2)
(0,5,0)
(1,3,0)
(0,3,2)
(0,1,3)
(1,1,1)
(0,5,1)
(0,7,0)
(2,1,0)
(1,5,0)
(1,3,1)
(0,1,4)
(1,1,2)
(0,3,3)
(1,1,2)
(0,5,2)
(0,9,0)
606
1783
1859
2930
2947
3026
3120
4076
4099
4150
4187
4273
4391
5175
5203
5243
5279
5336
5377
5431
5532
5670
0.1936
0.7056
0.153
0.007
0.546
0.484
0.425
0.2704
0.7056
0.068
(
.
(
2,0,0)
. .
1,0,2)
4655
4847
0.0144
0.007
0.017
a
1
a
Normalized to this value.
(
(
(
(
(
1,0,2)
0,4,2)
0,8,0)
2,2,0)
0,2,4)
In addition, Tables 4–6 give the Franck–Condon Fac-
tors (FCFs) corresponding to the direct double ionisa-
tion HCN SS HCN(X R ). Only FCFs for the
2
+
1
+
a
(2,0,1)
1
(
(
1,4,1)
0,0,5)
Table 5
(1,2,2)
(1,0,3)
a
2
+ 1
2+ 1
ꢀ1
Rovibrational levels of HCN ( D) and HCN ( P) (in cm ) using
1
0
the A component of these electronic states (without considering the
Renner–Teller interaction). We are also giving the Franck–Condon
factors (FCF) corresponding to the double ionisation from
Normalized to this value.
1
+
ꢀ1
HCN(X R ). See text for more details
levels located <6000 cm relative to the vibrational
zero point energies (ZPE) are given, since the PEFs cal-
culated here are not valid above this energy. For the
HCN(X R ) state we have used our own calculated
PEF.
The analysis of the FCFs shows that long progres-
sions in CN stretching, but insignificant CH excitation
should be observed experimentally, despite the fact that
both CH and CN bonds are longer in HCN than in
HCN neutral (cf. Table 2). There is considerable CN
extension in the p states resulting in long progressions
with very weak HCN (0,0,0)
transitions, making unambiguous determination of the
double ionisation potentials very difficult. For instance,
the first strong line in both HCN and DCN spectra (near
37.5 eV) is (at least) a doublet and the third strong line is
a triplet. Because the calculated FCF is so small, the first
significant peak in the spectra at 37.49 ± 0.02 (HCN)
and 37.52 ± 0.02 (DCN) is probably not the (0,0,0)–
2
+ 1
2+ 1
HCN ( P)
HCN ( D)
ꢀ1
ꢀ1
(
v
1
,v
2
,v
3
)
Energy (cm
)
FCF (v
1
,v
2
3
,v )
Energy (cm )
FCF
1
+
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
0,0,0)
0,2,0)
0,0,1)
0,4,0)
1,0,0)
0,2,1)
0,0,2)
0,4,1)
0,0,6)
1,0,1)
1,2,0)
0,0,3)
0,2,2)
1,4,0)
2,0,0)
0,1,2)
1,2,1)
0,2,3)
1,0,2)
0,0,4)
1,6,0)
0,0,4)
a
0
1152
1258
2304
2333
2420
2495
3447
3452
3568
3573
3664
3726
4569
4577
4586
4694
4713
4769
4817
4893
4972
0.020 (0,0,0)
(0,2,0)
0.183 (0,0,1)
(0,4,0)
0.009 (1,0,0)
0.009 (0,2,1)
0.510 (0,0,2)
(1,2,0)
0
972
0.261
0.0049
0.740
1358
1905
1931
2297
2743
2802
2882
3267
3337
3670
3750
3852
3890
4147
4228
4294
0.216
0.0049
2+
a
1
0.0541
ꢀ
2
(0,6,0)
2
+
1 +
0.580 (0,4,1)
0.009 (1,0,1)
0.383 (1,2,1)
(1,4,0)
(2,0,0)
0.111 (0,8,0)
(0,0,3)
HCN X R (0,0,0)
0.019
0.002
0.078
1
(1,2,1)
(0,6,1)
. . .
(0,0,4)
(3,0,0)
(1,0,4)
0.005
a
1
(0,0,0) transition. (This is the lower energy component
5554
5871
6544
0.394
0.589
0.044
of the doublet). This effect should be less strong in the
ꢀ
1
ꢀ1
ꢀ1
r
p
states. For energies >2000 cm
above the
Normalized to this value.
ZPE, the overlap between the PEFs of HCN X and

M. Hochlaf et al. / Chemical Physics 309 (2005) 291–301
299
2
+
1
those of HCN together with the anharmonic reso-
nances between the CN and CH stretching modes, give
intense FCFs for the combination modes such as
These may be due to the P state, which should lie in
the 40–41 eV energy range. The spectra of Fig. 2 are
quite difficult to analyse even using our highly accurate
ab initio calculations mainly because of ambiguity in the
(
v ,0,v ) (see Tables 4–6 for more details).
1
3
Tables 7 and 8 give the calculated rovibrational spec-
position of the (0,0,0)
(0,0,0) transitions (even using
2
+
3
ꢀ
1
1 + 1
tra of DCN (X R , D, R , P). The isotopic shifts
isotopic shifts) and the overlap between all the vibra-
tional progressions. Fig. 7 gives a tentative partial simu-
lation of the experimental spectrum using our computed
energies of the rovibrational levels and FCFs (cf. Tables
4–6) and assuming 50 meV experimental resolution. The
double ionisation energies can not be computed accu-
rately (typically within ± 0.1 eV even with the most accu-
rate theoretical methods as the ones used in the present
ꢀ
1
are calculated to be in the ranges of ꢁ500–520 cm
ꢀ
1
ꢀ1
for m , ꢁ100–120 cm for m and ꢁ60–70 cm for m ,
1
2
3
values which are consistent with those observed for the
+
neutral DCN and DCN cation [11,34]. Since the FCFs
correspond mostly to the excitation of the CN stretching
mode (m ) and the isotopic shift for this mode is smaller
3
than our experimental resolution, the TOF-PEPECO
spectra of HCN and DCN have similar shapes with rel-
atively few differences (cf. Fig. 2).
2
+
3
ꢀ
work). Here, the HCN
X R (0,0,0)
HCN
1
+
X R (0,0,0) transition was put at 37.2662 eV and we
have used the T values of Table 2 to position the other
e
excited states. We emphasize that these simulations are
not unique, despite the good accord between the simu-
lated and the measured spectra. To go further we need
at least calculations of all the rovibronic levels with
inclusion of all the couplings (such as Renner–Teller
and vibronic interactions), and of the FCFs for all the
electronic states. Better resolved experimental spectra
will also be needed. Nevertheless, the present compari-
son shows good agreement on vibrational spacings,
5
. Discussion
According to these calculations, only the X R , P,
3
ꢀ
3
1
1
+
2+
D and R states of HCN are likely to be seen in
the experimental data below 40 eV (see Table 2). The
spectra show resolved lines, on a continuum which
grows stronger at higher energy. At 48.4 eV photon en-
ergy, some new peaks appear at 39.5 eV and above.
Table 7
2+
3
ꢀ
2+ 1 +
Rovibrational levels of DCN (X R ) and DCN ( R )
3
ꢀ
1
+
X R
R
+
+
r
(J = 0)
,v ,v
p (J = 1)
(v ,v ,v
r
(J = 0)
,v ,v
p (J = 1)
(v ,v ,v
ꢀ1
ꢀ1
ꢀ1
ꢀ1
Energy (cm )
(
v
1
2
3
)
Energy (cm
)
1
2
3
)
Energy (cm
)
(v
1
2
3
)
Energy (cm
)
1
2
3
)
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
0,0,0)
0,2,0)
0,0,1)
0,4,0)
1,0,0)
0,2,1)
0,6,0)
0,0,2)
1,2,0)
0,4,1)
1,0,1)
0,8,0)
0,2,2)
1,4,0)
2,0,0)
0,0,3)
0,6,1)
1,2,1)
0,10,0)
0,4,2)
1,6,0)
2,0,1)
2,2,0)
0,2,3)
0,8,1)
1,4,1)
0,12,0)
0,0,4)
1,0,2)
0
803
(0,1,0)
(0,3,0)
(0,1,1)
(0,5,0)
(1,1,0)
(0,3,1)
(0,7,0)
(0,1,2)
(1,3,0)
(0,5,1)
(1,1,1)
(0,9,0)
(0,3,2)
(1,5,0)
(2,1,0)
(0,1,3)
(0,7,1)
(1,3,1)
405
1204
1623
1996
2196
2433
2782
2821
2968
3233
3396
3562
3639
3735
3949
4000
4026
4180
4337
4446
4497
4579
4698
4811
4825
(0,0,0)
(0,2,0)
(0,0,1)
(1,0,0)
(0,4,0)
(0,2,1)
(0,0,2)
(1,2,0)
(0,6,0)
(1,0,1)
(0,4,1)
(0,2,2)
(0,0,3)
(2,0,0)
(1,4,0)
(0,8,0)
(1,2,1)
(0,6,1)
(1,0,2)
(0,4,2)
(0,2,3)
(2,2,0)
(0,0,4)
(1,6,0)
(0,8,1)
0
956
(0,1,0)
(0,3,0)
(0,1,1)
(1,1,0)
(0,5,0)
(0,3,1)
(0,1,2)
(1,3,0)
(0,7,0)
(1,1,1)
(0,5,1)
(0,3,2)
(0,1,3)
(2,1,0)
(1,5,0)
(0,9,0)
(1,3,1)
(0,5,2)
(1,1,2)
(0,3,3)
479
1436
1650
2271
2391
2611
2799
3202
3345
3421
3570
3765
3928
3994
4132
4298
4359
4525
4553
4726
1212
1599
1804
2027
2388
2405
2581
2832
2998
3170
3229
3351
3570
3580
3628
3787
3948
4041
4115
4176
4322
4412
4417
4567
4720
4739
4754
1166
1804
1911
2128
2311
2734
2866
2950
3088
3280
3436
3539
3665
3820
3888
4045
4077
4243
4411
4446
4543
4595
4671
4771
4825
(0,11,0)
(0,5,2)
(1,7,0)
(2,1,1)
(2,3,0)
(0,9,1)
(0,3,3)
(0,10,0)
(1,4,1)

300
M. Hochlaf et al. / Chemical Physics 309 (2005) 291–301
Table 8
Rovibrational levels of DCN ( D) and DCN ( P) (in cm ) using
similar approach for HCN ( D) and HCN ( P)
lines at 38–38.75 eV must be blends of lines from the
2
+ 1
2+ 1
2+ 1
ꢀ1
1
lower states with a dominant contribution of D. Above
2
+ 1
1 +
3
8.8 eV the main contributor is R . Finally, we cannot
2
+ 1
2+ 1
DCN ( P)
DCN ( D)
exclude the excitation of odd and/or even quanta of
bending in the experimental spectra specially when
anharmonic and/or vibronic and/or Renner–Teller
interactions are involved, as we have already observed
such excitation in the simple photoionisation of tri-
atomic molecules [35–39].
ꢀ1
ꢀ1
Energy (cm )
(
v
1
,v
2
,v
3
)
Energy (cm
)
(v ,v ,v
1
2
3
)
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
0,0,0)
0,2,0)
0,0,1)
1,0,0)
0,4,0)
0,2,1)
0,0,2)
1,2,0)
0,6,0)
1,0,1)
0,4,1)
0,2,2)
0,0,3)
2,0,0)
1,4,0)
0,8,0)
1,2,1)
0,6,1)
2,0,1)
0,4,2)
2,2,0)
0,2,3)
1,6,0)
0
916
(0,0,0)
(0,2,0)
(0,0,1)
(0,4,0)
(0,2,1)
(1,0,0)
(0,0,2)
(0,6,0)
(1,2,0)
(0,4,1)
(0,2,2)
(0,0,3)
(0,8,0)
(1,0,2)
(0,6,1)
(1,2,1)
(0,2,3)
(2,0,0)
(0,4,2)
(0,8,0)
(1,2,2)
(0,0,4)
0
805
1194
1801
1838
2117
2364
2689
2764
2971
3044
3292
3511
3536
3584
3693
3867
3975
4120
4224
4398
4443
4484
4623
4636
4691
1036
1575
1769
1824
2108
2314
2529
2602
2809
2862
3071
3216
3310
3373
3528
3586
3625
3885
3914
3963
We turn now to the analysis of the PEPIPICO spectra
(Fig. 4), in light of the potential energy curve calcula-
tions. In Figs. 8 and 9, we show how the singlet and tri-
2+
plet electronic states of HCN lying in the 0–4 eV
internal energy range evolve along the RCH and RCN
coordinates. The triplets cross the singlets at several
RCH and RCN distances thus allowing spin-orbit predis-
sociation processes to occur. The spin–orbit selection
rules are relaxed for bent geometries so that the conver-
2+
sion of the HCN ions from one multiplicity to the
other is feasible leading to the dissociation products.
By examining the curve of Figs. 8 and 9, one can see that
breaking the CH bond should be much easier than
3
ꢀ
breaking CN, since the potential barrier of the X R
along the RCH coordinate (ꢁ1.67 eV) is smaller that
along RCN (ꢁ2.31 eV). The height of this barrier may
be less for bent structures of the dication, which is con-
sistent with the estimation of the PEPIPICO barrier
height of ꢁ1.3 ± 0.2 eV {i.e., 38.5 ± 0.2 eV (Section
0,10,0)
0,0,4)
1,0,2)
2
+
3
ꢀ
confirming the dominance of the CN stretching mode
excitation during the double photoionisation of HCN.
The spacings calculated between the higher levels of
4.2)
–
37.2 (IP of HCN
X R )}. The
2
+
+
+
HCN !H + CN reaction is expected to take place
3
00
at least along the lowest A PEFs resulting from the
3
ꢀ 3
2+
CN progressions (v P 3) agree well with the observed
X R – P conical intersection. The metastable HCN
ions observed experimentally are probably formed in
the low-lying vibrational levels of the X R and/or D
3
spacings of the resolved lines. Moreover, these simula-
tions suggest that the lines up to about 38 eV probably
arise from the two lowest electronic states. Intermediate
3
ꢀ
1
1
+
and/or R electronic states.
+
+
3
-
1
Π
3
1
∆
1
Σ
+
HCN
X Σ
Π
4
1
∆
Σ
1
1
+
Π
3
2
1
0
1
1
Π
1
+
1
3
-
1 Σ 1 ∆
X Σ
3
-
3
X Σ
1 Π
RCH 2.53
++
[
H
C
N]
3
7.0
37.5
38.0
38.5
39.0
39.5
40.0
40.5
Ionisation energy (eV)
2
3
4
5
6
Fig. 7. HCN TOF-PEPECO spectrum of Fig. 2 together with the
RCH / bohr
calculated spectra (see text for more details). We have added ꢁ50 meV
1
to the P T
e
value (cf. Table 1) in order to match better the TOF-
Fig. 8. Interaction between the singlets and the triplets along the RCH
distance. The RCN distance is set to 2.53 bohr.
1
PEPIPICO spectrum for the P state.

M. Hochlaf et al. / Chemical Physics 309 (2005) 291–301
301
[3] A. Marten, D. Gautier, T. Owen, D.B. Sanders, H.E. Matthews,
S.K. Atreya, R.P.J. Tilanus, J.R. Deane, Astrophys. J. 406
4
3
2
1
0
(
1993) 285.
[4] Q. Li, D.J. Jacob, I. Bey, R.M. Yantosca, Geophys. Res. Lett. 27
2000) 357.
1
1
Π
1
+
1
Σ
(
3
[5] B. Vinesland, E.E. Conn, C.J. Knowles, J. Westley, F. Wissing,
Cyanide in Biology, Academic Press, 1981, pp. 517–541.
[6] B.P. Mathur, L.E. Abbey, E.M. Burgess, T.F. Moran, Org. Mass
Spectrom. 15 (1980) 312.
1 Π
1
1
∆
[7] J.H.D. Eland, Chem. Phys. Lett. 203 (1993) 353.
[
8] M. Hochlaf, R.I. Hall, F. Penent, H. Kjeldsen, P. Lablanquie,
M. Lavoll e´ e, J.H.D. Eland, Chem. Phys. 207 (1996) 159.
9] J.H.D. Eland, Chem. Phys. 294 (2003) 171.
3
-
X Σ
[
[
10] T. Kinugawa, P. Lablanquie, F. Penent, J. Palaudoux, J.H.D.
Eland, J. Electron Spectrosc., in press.
2
.205 RCN
++
[
[
11] C. Fridh, L. Asbrink, J. Electron Spectrosc. 7 (1975) 119.
12] H. Koppel, L.S. Cederbaum, W. Domke, W. von Niessen, Chem.
Phys. 37 (1979) 303.
13] L.S. Cederbaum, H. Koppel, W. Domcke, Int. J. Quantum Chem.
15 (1981) 251.
[H
C
N]
2
3
4
5
6
[
[
[
[
RCN / bohr
14] J.H.D. Eland, T. Field, P. Baltzer, D.M. Hirst, Chem. Phys. 229
Fig. 9. Interaction between the singlets and the triplets along the RCN
distance. The RCH distance is set to 2.205 bohr.
(
1998) 149.
15] J.H.D. Eland, M. Hochlaf, G.C. King, P.S. Kreynin, R.J. LeRoy,
I.R. McNab, J.-M. Robbe, J. Phys. B 37 (2004) 3197.
16] M. Hochlaf, G. Chambaud, P. Rosmus, J. Chem. Phys. 108
6
. Conclusion
(
1998) 4047;
M. Hochlaf, R.I. Hall, F. Penent, J.H.D. Eland, P. Lablanquie,
Chem. Phys. 234 (1998) 249.
17] M. Hochlaf, F.R. Bennett, G. Chambaud, P. Rosmus, J. Phys. B
Our experimental resolution has allowed the observa-
2
+
tion of vibrationally resolved spectra of HCN with
very complex structures due to vibronic and Renner–
Teller interactions. The comparison between the experi-
mental and calculated spectra showed that the observed
structures are mainly due to the CN mode excitations
and to combination modes involving minor contribu-
tions from the CH stretch, despite the longer RCH and
RCN distances in the dication relative to the neutral
[
31 (1998) 2163.
[
[
18] M. Hochlaf, J.H.D. Eland, J. Chem. Phys. 120 (2004) 6449.
19] J.H.D. Eland, O. Vieuxmaire, T. Kinugawa, P. Lablanquie, R.I.
Hall, F. Penent, Phys. Rev. Lett. 90 (2003) 053003.
20] Y. Hikosaka, J.H.D. Eland, Chem. Phys. 299 (2004) 147.
21] P.J. Knowles, H.-J. Werner, Chem. Phys. Lett. 115 (1985)
259.
22] H.-J. Werner, P.J. Knowles, J. Chem. Phys. 89 (1988) 5803.
[23] P.J. Knowles, H.-J. Werner, Chem. Phys. Lett. 145 (1988) 514.
[24] T.H. Dunning, J. Chem. Phys. 90 (1989) 1007.
[
[
[
1
+
HCN X R . We stress the difficulty in this case of locat-
2
+
ing the HCN (0,0,0)
HCN(0,0,0) transitions be-
[
25] MOLPRO is a Package of Ab Initio Programs written by H.J.
cause of relatively small FCFs for the direct double
ionisation. Metastable HCN dications are found to
be formed in the low lying ro-vibrational levels of
Werner and P.J. Knowles. Further details at www.tc.bham.ac.uk/
molpro.
26] J. Senekowitsch, thesis of the University of Frankfurt, Germany,
1988.
[27] I.M. Mills, in: K.N. Rao, C.W. Mathews (Eds.), Molecular
Spectroscopy: Modern Research, Academic Press, 1972.
2
+
[
3
ꢀ
1
1 +
X R and/or D and/or R electronic states. The effi-
2
+
+
+
cient HCN !H + CN dissociation occurring for
internal energies greater than ꢁ1.3 ± 0.2 eV takes place
[
28] S. Carter, N.C. Handy, Comput. Phys. Rev. 5 (1987) 117.
3
00
along the lowest A PEFs and probably involves vib-
ronic, Renner–Teller and spin–orbit couplings.
[29] G. Herzberg, in: Molecular Spectra and Molecular Structure III,
Electronic Spectra and Electronic Structure of Polyatomic Mol-
ecules, Krieger, Malabar, 1991.
[
30] F. Penent, J.P. Grouard, J.L. Montmagnon, R.I. Hall, J. Phys. B
3 (1990) 2105.
[31] T.A. Field, J.H.D. Eland, Chem. Phys. Lett. 211 (1993) 436.
2
Acknowledgements
[
[
32] J.H.D. Eland, Mol. Phys. 61 (1987) 725.
33] M. Hochlaf, J. Palaudoux, A. Ben Houria, Transworld Research
Network (2004), in press.
34] webbook.nist.gov.
35] J. Liu, W. Chen, C.-W. Hsu, M. Hochlaf, M. Evans, S. Stimson,
C.-Y. Ng, J. Chem. Phys. 112 (2000) 10767.
M.H. thanks the NERSC (UC Berkeley, USA) for
computational time. We acknowledge the financial sup-
port of the EPSRC for the experimental part of the work.
[
[
[
[
36] J. Liu, M. Hochlaf, C.Y. Ng, J. Chem. Phys. 113 (2000)
7
988.
References
37] J. Liu, M. Hochlaf, G. Chambaud, P. Rosmus, C.-Y. Ng, J. Phys.
Chem. A 105 (2000) 2183.
[
[
1] A.G. Maki, J. Phys. Chem. Ref. Data 3 (1974) 221.
2] M. Banaszkrewicz, L.M. Lara, R. Rodrigo, J.J. Lopez-Moreno,
G.J. Molina-Cuberos, Icarus 147 (2000) 386.
[38] W. Chen, M. Hochlaf, P. Rosmus, G.Z. He, C.-Y. Ng, J. Chem.
Phys. 116 (2002) 5612.
[39] J. Liu, M. Hochlaf, C.-Y. Ng, J. Chem. Phys. 118 (2003) 4487.