Thermal reactivity of HNCO with water ice: An infrared and theoretical study

Article abstract of DOI:10.1016/S0301-0104(03)00024-7

The structure and energy of the 1:1 complexes between isocyanic acid (HNCO) and H₂O are investigated using FTIR matrix isolation spectroscopy and quantum calculations at the MP2/6-31G(d,p) level. Calculations yield two stable complexes. The first and most stable one (ΔE = 23.3 kJ/mol) corresponds a form which involves a hydrogen bond between the acid hydrogen of HNCO and the oxygen of water. The second form involves a hydrogen bond between the terminal oxygen of HNCO and hydrogen of water. In an argon matrix at 10 K, only the first form is observed. Adsorption on amorphous ice water at 10 K shows the formation of only one adsorption site between HNCO and ice. It is comparable to the complex observed in matrix and involves an interaction with the dangling oxygen site of ice. Modeling using computer code indicates the formation of polymeric structure on ice surface. Warming of HNCO, adsorbed on H₂O ice film or co-deposited with H₂O samples above 110 K, induces the formation of isocyanate ion (OCN^-) characterized by its v_asNCO infrared absorption band near 2170 cm^-1. OCN^- can be produced by purely solvation-induced HNCO dissociative ionization. The transition state of this process is calculated 42 kJ/mol above the initial state, using the ONIOM model in B3LYP/6-31g(d,p).

Full text of DOI:10.1016/S0301-0104(03)00024-7

Chemical Physics 288 (2003) 197–210
www.elsevier.com/locate/chemphys
Thermal reactivity of HNCO with water ice:
an infrared and theoretical study
1
Sebastien Raunier , Thierry Chiavassa, Alain Allouche, Francis Marinelli,
ꢀ
Jean-Pierre Aycard*
Physique des Interactions Ioniques et Moleꢀculaires, Equipe Spectromeꢀtries et Dynamique moleꢀculaire, UMR 6633,
^
Universiteꢀde Provence, Centre de Saint Jeꢀro^me, Boıte 252, F-13397 Marseille cedex 20, France
Received 30 October 2002; in final form 15 January 2003
Abstract
The structure and energy of the 1:1 complexes between isocyanic acid (HNCO) and H2O are investigated using
FTIR matrix isolation spectroscopy and quantum calculations at the MP2/6-31G(d,p) level. Calculations yield two
stable complexes. The first and most stable one (DE ¼ 23:3 kJ/mol) corresponds a form which involves a hydrogen bond
between the acid hydrogen of HNCO and the oxygen of water. The second form involves a hydrogen bond between the
terminal oxygen of HNCO and hydrogen of water. In an argon matrix at 10 K, only the first form is observed. Ad-
sorption on amorphous ice water at 10 K shows the formation of only one adsorption site between HNCO and ice. It is
comparable to the complex observed in matrix and involves an interaction with the dangling oxygen site of ice.
Modeling using computer code indicates the formation of polymeric structure on ice surface. Warming of HNCO,
adsorbed on H2O ice film or co-deposited with H2O samples above 110 K, induces the formation of isocyanate ion
(OCNÀ) characterized by its masNCO infrared absorption band near 2170 cmÀ1. OCNÀ can be produced by purely
solvation-induced HNCO dissociative ionization. The transition state of this process is calculated 42 kJ/mol above the
initial state, using the ONIOMmodel in B3LYP/6-31g(d,p).
ꢀ 2003 Elsevier Science B.V. All rights reserved.
1. Introduction
and reaction of atoms and molecules from the gas
on dust grains in the dense molecular clouds [2].
Present on the satellites of the outer planets and
on the comets, amorphous water ice is also be-
lieved to be the major constituent of interstellar
dust [3–6]. To understand the formation and
evolution of interstellar molecules, numerous
theoretical models with detailed physical and
chemical reaction pathways, have been developed
[7–10]. The tiny (micron and submicron) icy dust
grains present in cold molecular clouds prove to
More than 100 interstellar molecules have been
identified in the interstellar medium [1]. Many of
these molecules result from an efficient accretion
* Corresponding author.
E-mail addresses: sebastien.raunier@piimsdm.univ-mrs.fr
(S. Raunier), aycard@piimsdm3.univ-mrs.fr (J.-P. Aycard).
1
Also corresponding author. Tel.: 0491282705; fax:
0491288605.
0301-0104/03/$ - see front matter ꢀ 2003 Elsevier Science B.V. All rights reserved.
doi:10.1016/S0301-0104(03)00024-7

198
S. Raunier et al. / Chemical Physics 288 (2003) 197–210
be important catalysts. Heterogeneous reactions
are developed on the ice surface or in the icy
mantle of the grains [11] which are assumed to
play an important role in interstellar chemistry
[12]. For these reasons, we have been interested in
small molecules interactions with amorphous ice
surface [13–20].
to compare the experimental IR spectra with the
calculated ones. They are also instrumental in as-
signing observed absorptions, determining the
complex and the structure of absorption sites and
furthermore to modeling the reactivity of HNCO
on ice surfaces.
Since its discovery in 1979 by Soifer et al. [21]
in the protostar W33A, the 4.62 lm feature (2167
cmÀ1) has been extensively hunt up [22] and nu-
merous carriers, such as nitriles [23] and isonitr-
iles [24], have been proposed for this feature
called ‘‘XCN’’ band. The position and width of
this band led to assigning it to a solid molecular
species. In 1987, Grim and Greenberg [25] dis-
cussed the spectroscopic validity of the assign-
ment of this XCN band to nitriles and isonitriles
and proposed the XCN feature to the intense
asymmetric stretch of the isocyanate anion,
OCNÀ. Ten years later, Schutte and Greenberg
[26] detected three other weak bands of OCNÀ at
1296, 1206 and 630 cmÀ1 and concluded that
OCNÀ is a good candidate for the ‘‘XCN’’ band.
The behavior of the XCN feature when the ice
matrix was doped by electron donor or electron
acceptor molecules gave direct evidence that the
carrier is a negative ion [27,28]. Next, Hudson
et al. [29] confirm, from irradiated and photo-
lyzed ices, that the band produced in the labo-
ratory is due to OCNÀ. Novozamsky et al. [30]
proposed that the formation of OCNÀ is pre-
ceded by photochemical formation of isocyanic
acid, HNCO, followed by proton transfer to
some base such as NH3.
In this paper, we will be looking at the re-
activity of isocyanic acid, which has been de-
tected in the interstellar medium in gas phase
but never in the cold (10–100 K) interstellar
grains [31,32], on water ice surface or in bulk.
The purpose of this work is threefold: (1) to
obtain direct and accurate experiment results of
the HNCO . . . H2O complexes trapped in rare
gas cryogenic matrix, (2) to assess the chemical
stability of HNCO adsorbed on water ice sur-
faces, and (3) to determine the favorable condi-
tions for the OCNÀ formation.
2. Experimental section
Pure HNCO was synthesized using the method
described by Herzberg and Reid [33] and modified
by Sheludyakov et al. [34]. The isocyanic acid is
degassed before each deposition. Moreover, the
first fraction of isocyanic acid is evacuated. H2O
was degassed by successive freeze–thaw cycles
under vacuum before each use.
The apparatus and experimental techniques
used to obtain argon matrices have been de-
scribed elsewhere in the literature [35]. The rel-
ative concentrations Ar/H2O (50/1), Ar/HNCO
(700/1), Ar/H2O/HNCO (500/6/0.5) at room
temperature, were adjusted by standard mano-
metric techniques and sprayed onto a golden
copper plate cooled at 20 K held at a constant
pressure of 10À7 mbar and a constant deposition
rate (2 mmol/h). To study the reactivity of
HNCO with H2O ice, two types of experiments
were carried out. Gaseous HNCO (ꢀ10À6 mol)
was deposited at 10 K on a H2O amorphous ice
film with a 0.1 lm thickness previously prepared.
The techniques used to obtain amorphous ice
films from water/argon (1/50) gaseous mixture
have been described previously [15,19]. The de-
posited ice film thickness is deduced from the
calibration of infrared absorbance versus film
using optical interferences [36,37].
The behavior of HNCO was also studied when
it was co-deposited with H2O at 10 K, in a ratio 1/
10. The thermal activation of the samples was re-
alized by a progressive warming of the mirror
(from 10 K to near of 180 K), with a heating rate
of 0.9 K/min.
A Fourier Transform Infrared Spectrometer
(Nicolet series II Magna system 750) was used
to record the spectra samples in the range 4000–
400 cmÀ1 with a resolution of 0.12 cmÀ1 for
matrix experiments and 0.5 cmÀ1 for solid.
Experiments were monitored by FT-IR spec-
troscopy. Quantum calculations were undertaken

S. Raunier et al. / Chemical Physics 288 (2003) 197–210
199
3. Experimental results
complex. In the dOH region, a weak band at 1590
cmÀ1 is also relative to this species. The strong shift,
observed for the m1 mode of HNCO, suggests the
formation of a 1:1 complex where HNCO interacts
with H2O via its NH bond. Further, geometry and
frequency calculations will be performed to inves-
tigate the type of formed complexes.
3.1. Study of the HNCO . . . H2O complexes: cryo-
genic matrix experiments
Argon matrixes containing only HNCO or H2O
were prepared, yielding infrared absorptions that
were similar to those previously reported. Table 1
summarizes the observed absorptions with spectral
assignments based on the works of Ayers [38,39],
Couturier-Tamburelli et al. [40] for H2O, and Teles
et al. [41] for HNCO. Prior to the absorption study,
matrix isolation experiments were conducted with
different compound concentrations. The charac-
teristic absorption bands of free or multimers H2O
and HNCO multimers were well identified (Fig. 1)
from these experiments. The spectra recorded after
different co-depositions of Ar/HNCO/H2O at 20 K
show new absorption bands with respect to the
spectra of pure H2O and pure HNCO trapped in
argon matrix (cf. Fig. 1(a) and Table 1). The two
more intense vibrational bands located at 3296 and
2272 cmÀ1, shifted, respectively, by 215 cmÀ1 to the
lower frequencies and 13 cmÀ1 to the higher fre-
quencies with respect to the m1 and m2 mode of
HNCO monomer. It can be attributed to the pres-
ence of a molecular complex. In addition, in the mOH
region, between 3800 and 3550 cmÀ1 (cf. Fig. 1), a
3.2. Adsorption of HNCO on ice surface: thermal
evolution
Considering results obtained in argon matrix,
we have studied the thermal behavior of HNCO
adsorbed on amorphous water ice surface.
Amorphous ice spectra have been extensively
described in the literature [3,42]. In addition to
the bulk OH stretching and bending modes at
3250 and 1600 cmÀ1, respectively, they display a
weak feature at 3695 cmÀ1 which is relative to the
dangling OH mode. This latter is generally used
to follow the adsorption of molecules on ice
surfaces. Here, the dangling OH is not perturbed
and this observation indicates that there is no
interaction between HNCO and OH group.
When HNCO is deposited at 10 K, no change in
the position of the infrared ice absorption bands
is observed. The line shape and the position of
the m2 mode of HNCO at 2252 cmÀ1, the only
band of the HNCO IR spectrum that can be
observed, is the same that observed in the spec-
trum of solid isocyanic acid and similar to the
new weak absorption band observed at 3722 cmÀ1
,
17 cmÀ1 lower than the value observed for free m3
mode of H2O (3739 cmÀ1), is also attributed to the
Table 1
Experimental and calculated frequency shifts (in cmÀ1) for HNCO . . . H2O complexes (Dm ¼ mcomplex À mmonomer
)
Species
Assignments
Experiment
Calculation (MP2/6-31G(d,p))
Monomer Complex
Monomer HOw complex
OHw complex
m
m
Dmexp
)17
m
m
Dmcalc
m
Dmcalc
H2O
m3
m1
m2
m1
m2
m3
m4
m6
m5
3739
3722 (1)
4028
4013 (6.6)
3881 (1.6)
1674 (7.1)
)15
)8
)8
4024 (20)
)4
5
mOH asym
mOH sym.
dHOH sym.
mNH
mNCO asym.
mNCO sym.
dHNC
3639
1589
3511
2259
1316
770
3889
1682
3791
2366
1310
794
3894 (8.7)
1697 (12.6)
3799 (34.6)
2377 (100)
1320 (0.4)
1590 (6)
1
15
8
HNCO
3296 (52) )215
2272 (100)
3536 (100) )255
2375 (45.3)
1315 (1)
13
9
11
10
5
898 (33)
624 (0.3)
559 (7.3)
104
14
0
763 (45.4) )31
697
573
610
559
606 (0.1)
563 (14.5)
)4
4
dNCO
dNCO
575 (10)
2
Dm ¼ mðcomplexÞ À mðmonomerÞ; normalized intensities are in parenthesis.

200
S. Raunier et al. / Chemical Physics 288 (2003) 197–210
Fig. 1. (a) Infrared spectrum of HNCO and H2O trapped in argon matrix (0.5/6/600). (b) Infrared spectrum of HNCO trapped in
argon matrix (1/700). (c) Infrared spectrum of H2O trapped in argon matrix (1/50).
one published by Lowenthal et al. [43]. No
change is observed in the spectrum when the
sample is heated (0.9 K/min) until a temperature
of 110 K is reached. Then, a weak new absorp-
tion band appears at 2170 cmÀ1 (cf. Fig. 2) which
match very well the frequency of the asymmetric
stretch of OCNÀ reported in the literature data
[11,43]. This observation is indicative of an ioni-
zation of HNCO molecule on ice surface as re-
cently reported by Lowenthal et al. [43]. However
we do not observe the formation of the other
weak bands mentioned in the introduction at
1296, 1206, and 630 cmÀ1 due to the weak
abundance of OCNÀ formed at the interface of
H2O film and solid HNCO. For the same rea-
sons, the very weak and broadening absorption
bands of H3Oþ [16], that can be assigned to the
formed counterion, are not detected with IR
spectroscopy owing to overlap of its features with
those of H2O.
Above 110 K, the dangling OH signal begins
to decrease in intensity, then disappears at 130 K.
The latter phenomenon is due to the collapse of
amorphous ice pores during its evolution in
crystalline ice. This signal is similar to that ob-
served during desorption experiments of ozone
[18] or cyanoacetylene [20] adsorbed on amor-
phous ice surfaces. These results are comparable
to those obtained in the matrix experiments re-
ported in the former section. They indicate that
HNCO principally interacts with the dangling ice
oxygen sites (in
a
similar way than in
H2O . . . HNCO complex) than with the dangling
ice H sites (like HNCO . . . H2O complex). This
arrangement can be favorable to a proton trans-
fer from the hydrogen of HNCO to the oxygen of
H2O.
In the following, we compare this result to the
one obtained from the co-deposition of HNCO/
H2O mixture (1/10) where H2O is in excess.

S. Raunier et al. / Chemical Physics 288 (2003) 197–210
201
Fig. 2. IR spectra of HNCO deposited on amorphous water ice at 10 K (b) then warmed up with heating rate of 0.9 K/min. (a) Pure ice
at 10 K, (c) 80 K, (d) 90 K, (e) 100 K, (f) 110 K, (g) 125 K, (h) 135 K, and (i) 145 K. The inserts refer to the difference spectra between
the spectrum of HNCO deposited on ice at the temperature T and that of pure ice at the same temperature.
3.3. Co-deposition of HNCO with H2O at low
temperature: thermal evolution
When the sample is heated, about 130 K all the
vibration bands of the HNCO:ðH2OÞn complexes
strongly decrease. In the masNCO region, only the
solid component at 2252 cmÀ1 expected for the
solid HNCO can be again observed. A new ab-
sorption band is observed near of 2172 cmÀ1. This
frequency is very closed to that expected for the
OCNÀ species and suggests as previously described
a full ionisation of the HNCO molecule. However
in the bulk, OCNÀ is formed in a greatest amount
When HNCO and H2O are co-deposited at 10
K on the cold window, we observe the presence of
two shoulders at 3591 and 2970 cmÀ1 which are
not observed on the IR spectra of the solid HNCO
and solid H2O (Fig. 3). These new spectral features
are likely associated with the HNCO:ðH2OÞn
complexes formation in the solid and are different
of the 1:1 HNCO:H2O molecular complex previ-
ously observed in matrix. These two bands can be
assigned to a combination mode and to the mNH
stretching mode of HNCO. The more intense
bands of H2O and HNCO, i.e. the mOH mode and
the masNCO mode are also downshifted by nearly
40 and 10 cmÀ1 with respect to the mode of solid
HNCO and solid H2O.
as shown by the intensities of band at 2172 cmÀ1
.
This band that appears here is slightly shifted but
more intense than in the experiment adsorption.
Although these conditions favour the OCNÀ for-
mation, we do not detect the other weak bands of
OCNÀ and the H3Oþ ion absorptions.
Holding the temperature at 130 K during sev-
eral minutes, whereas the vibrational bands of the

202
S. Raunier et al. / Chemical Physics 288 (2003) 197–210
Fig. 3. Co-deposition of HNCO and H2O, (a) pure HNCO 10 K, (b) pure H2O 10 K, (c) HNCO/H2O 10 K, (d) 50 K, (e) 110 K,
(f) 130 K, t ¼ 00, (g) 130 K, t ¼ 300, (h) 150 K. * associated to CO2.
HNCO:ðH2OÞn complexes goes on to decrease
then disappear, we observe both a slight increasing
of the m2 band of OCNÀ and the appearance of a
feature at 2295 cmÀ1 which is in good agreement
with the mCN stretching mode of cyanic acid,
HOCN, reported in the literature [41]. This ob-
servation suggests the possibility of a new proton
experimental results. The experimental IR spectra
were compared with the calculated ones to assign
observed vibrational absorptions and determine
the structure of the complex in matrix. The struc-
tures of adsorption sites and the thermal ioniza-
tion of HNCO on ice surface were also modeled.
transfer between protonated ice, H3OþðH2OÞ and
4.1. Structure of the 1 . . . 1 complex
n
OCNÀ to lead at HOCN, isomer of the isocyanic
acid. Above 135 K, the HOCN, HNCO and
OCNÀ species are sublimated and only the water
ice vibrational bands are observed.
To establish the molecular structure of this
complex, ab initio calculations were carried out
with Gaussian 98 [44] at the MP2/6-31G(d,p) level
of theory [45]. Several arrangements are possible in
the H2O and HNCO subunits in the complexes.
Water and isocyanic acid, are both simultaneous
proton donor and acceptor compounds and are
expected to form hydrogen bonds to nitrogen and
oxygen, respectively. Four kinds of complexes are
thus considered (Fig. 4). The NHw form which
involves a hydrogen bonding between hydrogen of
water (Hw) with the nitrogen of isocyanic acid; the
4. Quantum calculations
Our experimental results point out to the for-
mation of a 1:1 complex in the matrix, adsorption
of HNCO on the dangling oxygen sites of the ice
surface and thermal formation of OCNÀ. Quan-
tum calculations were carried out to interpret our

S. Raunier et al. / Chemical Physics 288 (2003) 197–210
203
Fig. 4. Starting for theoretical calculations of the complexes.
HOw form, which involves an hydrogen bonding
between the hydrogen of isocyanic acid and the
oxygen of water (Ow); the OHw form, which in-
volves an hydrogen bonding between hydrogen of
water (Hw) and the oxygen of isocyanic acid, and
the T form, which involves an interaction between
the electrophilic carbon of isocyanic acid and the
nucleophilic oxygen of water.
Geometry of the complexes (cf. Table 2): From
the starting geometries in Fig. 4, all the geometri-
cal parameters were fully optimized using the
Berny optimization procedure [46]. The harmonic
vibrational frequencies were determined at the
stationary points. Calculations give only two local
minima with a Cs symmetry structure. The first
ꢁ
one, with a rO...H value of 2.177 A, corresponds to
the OHw structure. The second corresponds to the
HOw form and is stabilized by a strong hydrogen
ꢁ
bond ( rH...O ¼ 1:87 A). The NHw form and the T
shaped structure do not correspond to a local
minimum. The optimized structures of the com-
plexes with bond lengths are reported in Fig. 5 and
in Table 2.
Identification of the complex: The previous cal-
culation provides valuable insight into the stability
and the spectroscopic features of the complexes.
First, the complexing effects can be observed on
the geometry of the partner molecules (cf. Table 2)
and on their most significant stretching mode fre-
quencies (cf. Table 1). The HOw complex is pre-
dicted to be the most stable with respect to the
BSSE
Tot
BSSE
HNCO
BSSE
H2O
monomers (DEðE
À E
À E
Þ ¼ À28:3
kJ/mol). The energy difference, )23.3 kJ/mol, with

204
S. Raunier et al. / Chemical Physics 288 (2003) 197–210
Fig. 5. Optimized structures of HNCO . . . H2O complexes (Cs symmetry).
respect to the OHw form, is significant at this level
of theory and can explain the lack of the second
structure. By comparing the m1 (NH stretch) and m2
(NCO antisymmetric stretch) vibrational fre-
quency shifts of the complex compared to those
calculated for the free HNCO, we are able to
identify the form present in argon matrix (cf. Table
1). The predicted vibrational frequencies for the
OHw complex are slightly perturbed in comparison
to the values of the respective monomers. This is
not the case of the HOw complex where the m1
mode of HNCO is very sensitive to the interaction
with H2O. Experimentally, this mode is down-
shifted by 215 cmÀ1 and this value is in good
agreement with the calculated oneÕs (255 cmÀ1).
The comparison of the experimental and theoret-
ical shifts for the m2 mode of HNCO and for the m1
and m3 modes of H2O complex confirms the pres-
ence of a HOw complex and the lack of the OHw
one. These results confirm the acidic character for
isocyanic acid.
known as P-ice [52–54]. The orbital basis set is the
standard 6-31g(d,p).
In the present section, we considered only the
unrelaxed ice surface. Hence, the optimization
process concerns only the adsorbed HNCO mole-
cules. Each (1 ꢁ 1) water surface unit cell contains
one isocyanic acid molecule. The automatic gra-
dient not being implemented in CRYSTAL98, the
equilibrium adsorption structures are calculated
by quadratic interpolation techniques [48]. The
PHF energies were corrected by the electron cor-
relation contribution evaluated by the GGA–DFT
functional [55].
Various positions were explored in order to
determine the nature of the adsorption sites. After
optimization only two stable structures were
found. Their total energies (in Hartree) are re-
ported in Table 3.
• For the first structure (Fig. 6(a)), HNCO is up-
right above the dangling O site. The N–H . . . O
ꢁ
distance (1.783 A) is characteristic of a strong
hydrogen bond. This structure coming from
the HOw complex observed in matrix is illus-
trated in Fig. 6(a), and it is noted HOiv (i for
ice and v for vertically HNCO).
4.2. Structure of the adsorbed layer
The Periodic Hartree–Fock (PHF) method was
successfully used in our earlier papers [16,47–50] to
model the reactivity of small molecules adsorbed
on amorphous ice; therefore it is also applied in
the present case. The methodology is implemented
in the CRYSTAL98 computer code [51]. As pre-
viously stated, the ice substrate is represented by a
slab of two molecular bilayers, periodic in two
dimensions, parallel to the [0 0 0 1] face and well
• The structure displayed in Fig. 6(b) shows the
geometry obtained for the second minimum
(the most stable, cf. Table 3). In this structure,
coming from a HOw complex (noted here
HOif , i for oxygen of ice and f for flattened
HNCO), the HNCO is flattened on the ice sur-
face and HNCO is stabilized by two strong hy-
drogen bonds. One between the isocyanic acid

S. Raunier et al. / Chemical Physics 288 (2003) 197–210
205
Table 3
Total energies of HNCO adsorption sites and total energies of dimers per water unit of ice
Adsorption sites
Dimers
Total energy (relative energy)b
Total energya (relative energy)b
a
HOiv
HOif
DA
DB
DC
HF
)776.0922
(11.8)
)776.0967
(0.0)
)943.7561
(18.9)
)943.7591
(11.0)
)943.7633
(0.0)
GGA
)779.7812
(23.6)
1.783
)779.7902
(0.0)
1.934
)948.2672
(28.3)
)948.2720
(15.7)
(2.004d
1.734e
)948.2780
(0.0)
2.616d
dc (O . . . H–O)
dc (O . . . H–N)
2.108d
1.734
1.783e
a Hartree.
b kJ=molÀ1
.
c
ꢁ
A from HF calculations.
d (O4 . . . H25–O).
e (O10 . . . H1–N2).
Fig. 6. Optimized site structures of isocyanic acid adsorbed on ice surface (CRYSTAL98): (a) HOiv (i for ice and v for vertically
HNCO); (b) HOif (i for ice and f for flattened HNCO).
and the dangling lone pair of electrons of a sur-
face oxygen atom and one between the dangling
OH and the carbonyl oxygen of HNCO. The as-
sociated N–H . . . Oi and O . . . H–Oi distances
gling hydrogen, but this site, corresponding to a
OHw complex, is less stable in total energy as
calculated for isolated complex (cf. Table 2). These
theoretical results show that on a proton ordered
ice surface, HNCO exhibits two stable structures
(HOiv and HOif ). The appreciable difference in
stability (23 kJ/mol) between the two structures
ꢁ
are 1.734 and 1.934 A, respectively.
We also tried to adsorb the molecule above a
point of low electronic density, i.e. above a dan-

206
S. Raunier et al. / Chemical Physics 288 (2003) 197–210
optimization and DC corresponds to the
ðH–NC@OÞ . . . ðH–NC@OÞ dimer. Where
might justify in the experience the absence of the
less stable one, HOiv. The HOif structure, the most
ꢁ
stable, displays a shorter distance (1.935 A) com-
pared with the value calculated for the OHw
2
ðH–NC@OÞ1 denotes the initial HNCO molecule
1
shown in Fig. 2.
ꢁ
complex (2.177 A). Thus, due to a stronger inter-
These respective optimized topologies are dis-
played in Fig. 7 with atomic numbering. The total
energies are reported in Table 3. The analysis rel-
ative to the DC system suggests that the formation
of adsorbate islands can largely transform the
initial HOi configuration.
action between HNCO and H2O, a change of the
vibrational frequency of the dangling OHÕs should
be observed following adsorption. However, when
HNCO is deposited at low temperature on the
H2O surface, the dangling OH is not perturbed.
Obviously, at this temperature, some solid aggre-
gates have been deposited. Due to their hydrogen
donor and acceptor properties, HNCO molecules
could form more compact complexes on the ice
surface. So, we think the interaction existing be-
tween HNCO molecules, can play a role on the
spatial orientation of individual molecule and on
the lack of perturbation of the OH group. We
consider here only distributions containing two
non-equivalent HNCO molecules per surface unit
cell. These molecules tend to form dimers. From
the precedent stable structure HOif three likely
HNCO associations were investigated:
For the two DA and DB dimer forms, the
ðHNCOÞ . . . H–O length are similar (2.106 and
1
ꢁ
ꢁ
2.004 A, respectively, instead of 1.934 A in HOi
model). The situation is different for the most
stable dimer structure DC. The consequence of
displacements of the second molecule in the vi-
cinity is to attenuate the ðHNC@O4Þ . . . ðH25–OÞ
hydrogen bond. This hydrogen bond length, esti-
ꢁ
mated at 2.616 A, is significantly longer than 1.934
ꢁ
A in the HOi model. The resulting form of this
penetration must be related to the formation of a
weak interaction between the N6 of ðHNCOÞ and
2
the ice dangling H25 hydrogen: dðN6–H25Þ ¼
ꢁ
ꢁ
3:07 A. The O4–H5 bond length is 2.251 A, cor-
responding to a weak hydrogen bond. These
changes could lead to further stabilization for the
DC form. Hence, the data in Table 3 show that the
dimerization has a great impact on the spatial
orientation of individual molecules. As a result
• In the first one, DA, the arrangement of the two
molecules of HNCO corresponds to the dimer
found in gaseous phase (Fig. 7).
• The other two associations DB and DC are as-
sumed to be found for the solid state: DB corre-
sponds to a ‘‘head to tail’’ form before
Fig. 7. Optimized structures of isocyanic acid dimers adsorbed on ice surface.

S. Raunier et al. / Chemical Physics 288 (2003) 197–210
207
two consequences must be drawn: (i) the second
molecule of HNCO breaks the former hydrogen
bond, O4–H25 and forms new hydrogen bonds
(O4–H5 and N6–H25), and (ii) the HOw structure
becomes favored.
This dimer formation on ice surface is consis-
tent with the lack of perturbation on the dangling
OH absorption in the IR spectra. In conclusion of
this paragraph, the best overall qualitative agree-
ment with experiment is obtained with the dimer
model.
structure of the system, no covalent bond had to
be cut, the two subsystems being linked together
only by van der Waals and hydrogen bonds. The
choice of a quantum level for the low-level system,
even if semi-empirical, ensures a better polariz-
ability of the larger cluster and a better represen-
tation of the embedding of the reactive part in the
entire cluster. From the former paragraph, it can
be deduced that the lack of perturbation of the
dangling OH IR band observed is due to the for-
mation of HNCO dimers or polymers and because
of these dimers, the isocyanate oxygen interacts
weakly with the dangling OH. Nevertheless, it can
be reasonably expected that the ionization process
concerns the ad-molecule that lies close to the
surface and not the other one. Therefore, the
present study involves one isocyanic acid molecule
only.
The whole system was optimized without any
restriction. In fine, we can observe in Fig. 8 that
the structure is very similar to the P-ice that
formed the starting point of the periodic calcula-
tion: the chair-shape units constituting the ice
structure are only slightly distorted but they re-
main quite recognizable. In good agreement with
the periodic calculation, the system stability is
4.3. Ionization of the isocyanic acid on ice
The periodic methods have proved to be very
efficient in determining the structure of the ad-
sorbed phases but they become very prohibitive
and unsuited in dealing with the system reactivity.
In this case, it is much easier to revert to the cluster
model such as the model we used in a former work
[16]. We therefore carried out the calculation on
the system constituted with HNCO and its four
nearest water molecule neighbors. Nevertheless,
the model used in reference [16] has been refined
by introducing 18 extra water molecules embed-
ding the small cluster. The whole system is calcu-
lated using Gaussian 98 [44] and the ONIOM
model [56] with B3LYP/6-31g(d,p) for the high-
level cluster and PM3 for the low-level. Due to the
ꢁ
ensured by a strong hydrogen bond (1.663 A)
(noted HAO1) with a dangling O of the H2O ice
surface. The ONIOMcalculation also confirms
Fig. 8. Adsorbtion of isocyanic acid (a) and ionized isocyanic acid (b) on relaxed ice surface (ONIOMcluster approximation).

208
S. Raunier et al. / Chemical Physics 288 (2003) 197–210
À
þ
ꢁ
that the oxygen isocyanic acid interacts to a lesser
extent with the ice dangling OHs since the corre-
OCN . . . H–O H2 of 1.571 A (noted OHB). The
isocyanate oxygen is bound to the ice surface by
ꢁ
sponding distances are 2.123 and 2.181 A (noted
OHB).
ꢁ
ꢁ
two H-bonds of 1.704 A (NHA) and 1.806 A, re-
spectively.
The isocyanic acid ionization on the ice surface
was modeled in scanning the NH (noted NHA)
distance, the remaining of the large cluster being
relaxed without further constraint. The resulting
potential energy surface is displayed in Fig. 9. This
The minimum in energy, corresponding to the
isocyanate ion adsorbed on ice, occurs for the
ꢁ
distance values NHA ¼ 1:704 A and HAO1 ¼
ꢁ
1:004 A. The gain in energy is provided mainly by
the contribution of two strong hydrogen bonds,
with the already mentioned dangling OH, and the
ꢁ
figure shows a maximum in energy at NH ꢀ 1:55 A.
ꢁ
Given the size and the complexity of the system, it
is difficult to automatically calculate the corre-
sponding stationary state or to make an IRC cal-
culation. However, we can reasonably consider
that the maximum of Fig. 9 is a good approxi-
mation of the transition state of the ionization
reaction of isocyanic acid. The energy of this point
is 42 kJ/mol above the initial state. Due to the very
large number of variables, it has not been possible
to calculate the ZPE correction to this barrier. The
geometry corresponds to the capture of the proton
by the closest dangling oxygen on the ice surface.
In this transition state (no displayed in Fig. 8), the
H3Oþ counterion is located near the OCNÀ. The
bond lengths are 1.571 and 1.806 A. In the final
geometry, the hydronium ion H3Oþ is H-bonded
to the oxygen of OCNÀ, whereas in the transition
state it was bonded to N. The energy gain gener-
ated is 17 kJ/mol (cf. Fig. 8(b)). The net Mulliken
atomic charges are )0.52 electron on O, +0.19
electron on C and –0.46 electron on N. So the
electrostatic potential field is roughly the same on
the two ends of the moiety. The weak distance
between O of OCNÀ and H of the H3Oþ can be a
favourable condition to a new proton transfer to
OCNÀ. Leading to the cyanic acid, HOCN, as
previously reported in the experiment and corre-
sponding to the transfer of HB from O2 to O of
OCNÀ.
two being bonded by
a
hydrogen bond
Fig. 9. Potential energy surface corresponding to the isocyanic acid ionization process on ice surface, transfer of HA from N to O1.

S. Raunier et al. / Chemical Physics 288 (2003) 197–210
209
band at 2295 cmÀ1, so we can say that H2O solid
supports this isomerization process. Such as pro-
ton transfer on amorphous ice film have been al-
ready observed to explain the formation of
chloroformylketene from HCl and carbon subox-
ide (C3O2) [16]. Because of temperatures measured
in the interstellar cloud (T < 100 K) [57], this
study shows that the mechanism (ionization by
solvation) could not occur, since the observed
transition to OCNÀ begins above 100 K. In these
conditions, it appear more probable that the
OCNÀ ion observed by the ISO spectra from
its intense band at 4.62 lm results from an acid/
base mechanism operable at cloud temperatures
involving HNCO and NH3 as reported by Hudson
et al. [29].
5. Discussion and conclusion
The analysis of our results shows that HNCO
trapped in an argon matrix or adsorbed on an
amorphous ice surface, acts as an electrophil by
means of its acid hydrogen atom. For this reason,
only one kind of HOw complex (HOif adsorption
site, respectively) is obtained which is stabilized by
an hydrogen bond between the acid hydrogen
atom of HNCO and the oxygen atom of water (the
dangling O on the ice surface, respectively). From
theoretical calculations, a second minimum, less
stable, is obtained for a OHw complex (corre-
sponding to the dangling OH adsorption site, re-
spectively). This kind of complex stabilized by an
hydrogen bond between the water hydrogen and
the oxygen atom of isocyanic acid is not observed
experimentally. For the dangling O adsorption
site, the isocyanic acid is flattened on the ice sur-
face and complexed by an other HNCO molecule.
This polymer (HNCO . . . HNCO . . .) formation
process on ice surface is consistent with the lack of
perturbation on the dangling OH absorption, and
the line shape and position of the HNCO modes
are the same as those observed in solid isocyanic
acid [43] infrared spectrum.
Acknowledgements
The theoretical part of this work was conducted
ꢀ
with the technical means of ‘‘Centre Regional de
Competence en Modelisation moleculaire de
ꢀ
ꢀ
ꢀ
Marseille’’.
Above 110 K, the dangling OH begins to de-
crease in intensity, then disappears at 130 K. The
latter phenomenon is due to the collapse of the
amorphous ice pores during its evolution in crys-
talline ice. This study shows that HNCO is stable
on ice below 140 K. The thermal behavior of ad-
sorbed HNCO or trapped HNCO in the H2O so-
lid, shows the appearance of a new infrared
absorption band observed near 2170 cmÀ1 which is
attributed to the OCNÀ. Modeling of the thermal
isocyanic acid ionization process on the ice sur-
face, shows a low energy barrier of 42 kJ/mol
which can explain that the formation of OCNÀ
occurs at temperature about 130 K. OCNÀ is
produced by purely solvation-induced dissociative
ionization. The isocyanate ion adsorbed on ice, is
stabilized by two strong hydrogen bonds and in
the final geometry, the hydronium ion is H-bonded
to the oxygen of OCNÀ with an energy gain of 17
kJ/mol. In this last arrangement the hydronium
ion can transfer a proton to the OCNÀ ion to form
cyanic acid (HOCN) characterized by an intense
References
[1] G. Winnewisser, C. Kramer, Space Sci. Rev. 90 (1999)
181.
[2] S.B. Charnley, P. Ehrenfreund, Y.-J. Juan, Spectrochim.
Acta Part A 57 (2001) 685.
[3] V.F. Petrenko, R.W. Whitworth, in: Physics of Ice, Oxford
University Press, New York, 1999.
[4] M.S. Westlty, R.A. Baragiola, R.E. Johnson, G.A. Bar-
atta, Nature 373 (1995) 405.
[5] D. Blake, L. Allamendola, S.A. Sandford, D. Hudgins,
F. Freund, Science 254 (1991) 254, 548.
[6] S.A. Sandford, L. Allamendola, Icarus 76 (1988) 201.
[7] B.H. Andrew (Ed.), I.A.U. Symposium 87, Interstellar
Molecules, Reidel, Dordrecht, The Netherlands, 1980.
[8] C. Joblin, A. Jones (Eds.), Solid Interstellar Matter: The
ISO Revolution; Les Houches Workshop; L. dÕHende-
court, Springer/EDP Sciences, New York, 1999.
[9] I.W.M. Smith, B.R. Rowe, Acc. Chem. Res. 33 (2000) 261,
and references therein..
[10] C. Barrientos, P. Redondo, A. Largo, Recent Res. Dev.
Phys. Chem. 1 (1997) 51.
[11] (a) L. dÕHendecourt, M. Jourdain de Muizon, E. Dartois,
M. Breitfellner, P. Ehrenfreund, J. Benit, F. Boulanger,
J.L. Puget, H.J. Habing, Astron. Astrophys. 315 (1996)
315, L365;

210
S. Raunier et al. / Chemical Physics 288 (2003) 197–210
(b) F. Borget, T. Chiavassa, J.-P. Aycard, Chem. Phys.
Lett. 348 (2001) 425.
[32] B.E. Turner, R. Terzieva, E. Herbst, Astrophys. J. 518
(1999) 699.
[12] (a) I. Tamburelli, T. Chiavassa, F. Borget, J. Pourcin,
J. Phys. Chem. A 102 (1998) 423;
[33] G. Herzberg, C. Reid, Discuss. Faraday Soc. 9 (1950) 92.
[34] Y.L. Sheludyakov, F.Z. Shubareva, V.A. Golodov, Russ.
J. Appl. Chem. 67 (1994) 1081.
(b) J.D. Graham, J.T. Roberts, J. Phys. Chem. B 104
(2000) 978.
[35] N. Pietri, B. Jurca, M. Monnier, M. Hillebrand,
J.-P. Aycard, Spectrochim. Acta Part A 56 (2000) 157.
[36] M.A. Zondlo, A. Loewenshuss, C.J. Nielsen, J. Phys.
Chem. B 101 (1997) 8696.
[13] J.P. Devlin, J. Phys. Chem. 96 (1992) 6185.
[14] B. Rowland, N.S. Kadagathur, J.P. Devlin, J. Chem. Phys.
102 (1995) 13.
[15] I. Couturier-Tamburelli, T. Chiavassa, J. Pourcin, J. Phys.
Chem. B 103 (1999) 3677.
[37] A. Givan, A. Loewenshuss, C.J. Nielsen, J. Chem. Phys.
Lett. 275 (1997) 98.
[16] A. Allouche, I. Couturier-Tamburelli, T. Chiavassa, J.
Phys. Chem. B 104 (2000) 1497.
[38] G.P. Ayers, D.E. Pulling, Spectrochim. Acta Part A 32
(1976) 1629.
[17] F. Borget, T. Chiavassa, A. Allouche, J.-P. Aycard, J.
Phys. Chem. B 105 (2001) 449.
[39] G.P. Ayers, D.E. Pulling, Spectrochim. Acta Part A 32
(1976) 1695.
[18] C. Manca, A. Allouche, J. Chem. Phys. 114 (2001)
4226.
[40] I. Couturier-Tamburelli, T. Chiavassa, J.-P. Aycard,
J. Am. Chem. Soc. 121 (1999) 3756.
[19] F. Borget, T. Chiavassa, A. Allouche, F. Marinelli, J.-P.
Aycard, J. Am. Chem. Soc. 123 (2001) 10668.
[20] C. Manca, C. Martin, A. Allouche, P. Roubin, J. Phys.
Chem. B 105 (2001) 12861.
[41] J.H. Teles, G. Maier, B.A. Hess, L.J. Schaad, M. Winne-
wisser, B.P. Winnewisser, Chem. Ber. 122 (1989) 753.
[42] B. Rowland, J.P. Devlin, J. Chem. Phys. 94 (1991) 812.
[43] M.S. Lowenthal, R.K. Khanna, M.H. Moore, Spectro-
chim. Acta Part A 58 (2002) 73.
[21] B.T. Soifer, R.C. Puetter, R.W. Russel, S.P. Willner, P.M.
Harvey, F.C. Gillet, Astrophys. J. 232 (1979) L53.
[22] J.H. Lacy, F. Baas, L.J. Allamandola, S.E. Persson, P.J.
McGregor, C.J. Londsdale, T.R. Geballe, C.E.P. Van De
Bult, Astrophys. J. 523 (1984) 276.
[44] M.J. Frisch et al., Gaussian 98, Revision A.7, Gaussian
Inc., Pittsburgh, PA, 1998.
€
[45] C. Moller, M.S. Plesset, Phys. Rev. 46 (1934) 618.
[46] H.B. Schlegel, J. Comput. Chem. 3 (1984) 214.
[47] F. Marinelli, A. Allouche, Chem. Phys. 272 (2001) 137.
[48] A. Allouche, P. Verlaque, J. Pourcin, J. Phys. Chem. B 102
(1998) 89.
[23] H.P. Larson, D.S. David, J.H. Black, U. Frisk, Astrophys.
J. 299 (1985) 873.
[24] L. dÕHendecourt, L.J. Allamandola, R.J.A. Grim, J.M.
Greenberg, Astron. Astrophys. 158 (1986) 119.
[25] R.J.A. Grim, J.M. Greenberg, Astrophys. J. 321 (1987)
L91.
[49] A. Allouche, J. Phys. Chem. 100 (1996) 1820.
[50] A. Allouche, J. Phys. Chem. 100 (1996) 17915.
[51] V.R. Saunders, R. Dovesi, C. Roetti, M. Causa, N.M.
ꢂ
[26] W.A. Schutte, J.M. Greenberg, Astron. Astrophys. 317
(1997) L43.
Harrison, R. Orlando, C.M. Zicovitch-Wilson, CRYS-
ꢂ
TAL98 userÕs manual, Universitadi Torino, Italy.
[27] K. Demik, E. Dartois, L. dÕHendecourt, M. Jourdain de
Muizon, A.M. Heras, M. Breitfellner, Astron. Astrophys.
339 (1998) 553.
[52] C. Pisani, S. Casassa, P. Ugliengo, Chem. Phys. Lett. 1253
(1996) 201.
[53] S. Casassa, P. Ugliengo, C. Pisani, J. Chem. Phys. 106
(1997) 8030.
[28] R.L. Hudson, M.H. Moore, Astron. Astrophys. 357 (2000)
787.
[54] G. Bussolin, S. Casassa, C. Pisani, P. Ugliengo, J. Chem.
Phys. 108 (1998) 9516.
[29] R.L. Hudson, M.H. Moore, P.A. Gerakines, Astrophys.
J. 550 (2001) 1140.
[55] J.P. Perdew, Y. Wang, Phys. Rev. B 33 (1986) 8800.
[56] S. Dapprich, I. Komaromi, K.S. Byun, K. Morokuma,
M.J. Frisch, J. Mol. Struct. Theochem. 461 (1999) 1.
[57] L. dÕHendecourt, E. Dartois, Spectrochim. Acta Part A 57
(2001) 669.
[30] J.H. Novozamsky, W.A. Schutte, J.V. Keane, Astron.
Astrophys. 379 (2001) 588.
[31] E.F. van Dishoek, G.A. Blake, Annu. Rev. Astron.
Astrophys. 36 (1998) 317.