The microwave spectrum of cyanoformamide

Article abstract of DOI:10.1016/j.jms.2005.01.004

The spectrum of cyanoformamide, NCCONH₂, has been measured between 18 and 40 GHz using a Hewlett-Packard spectrometer with Stark modulation. The molecule is somewhat unstable and could most conveniently be measured in a flow system. The quadrupole hyperfine structure due to the two nitrogen nuclei has been analyzed for the ground state, and quadrupole coupling constants, rotational constants, and centrifugal distortion constants have been determined for the ground state. A rough determination of the components of the electric dipole moment was possible from the Stark shifts of suitable transitions.

Full text of DOI:10.1016/j.jms.2005.01.004

Journal of Molecular Spectroscopy 231 (2005) 131–136
www.elsevier.com/locate/jms
The microwave spectrum of cyanoformamide
1
Jørn J. Christiansen
Physikalisch-Chemisches Institut der Justus-Liebig-Universit a¨ t, Heinrich-Buff-Ring 58, D-35392 Giessen, Germany
Received 1 October 2004; in revised form 29 December 2004
Available online 19 February 2005
Abstract
The spectrum of cyanoformamide, NCCONH , has been measured between 18 and 40 GHz using a Hewlett–Packard spectrom-
2
eter with Stark modulation. The molecule is somewhat unstable and could most conveniently be measured in a flow system. The
quadrupole hyperfine structure due to the two nitrogen nuclei has been analyzed for the ground state, and quadrupole coupling
constants, rotational constants, and centrifugal distortion constants have been determined for the ground state. A rough determi-
nation of the components of the electric dipole moment was possible from the Stark shifts of suitable transitions.
ꢀ
2005 Elsevier Inc. All rights reserved.
Keywords: Prebiotic molecules; Quadrupole hyperfine coupling; Electric dipole moment
1
. Introduction
1992 [7]. However, the rotational spectrum of cyanofor-
mamide itself has never been reported. This spectrum as
measured in the microwave region is presented here,
including 126 assigned transitions for the ground state,
an analysis of the quadrupole hyperfine structure due
to the two nitrogen nuclei, a rotational and centrifugal
distortion analysis for the ground state data and a deter-
mination of the electric dipole moment components.
This work has been the starting point for measurements
in the millimeter wave region which will be published
shortly [8]. Assignments of data for several excited
vibrational states will be reported in a later publication
[9].
If cyanogen, NCCN, and water are made to react, the
product is cyanoformamide, NCCONH . This species
2
had been known since 1870, when it was synthesized
by Beketoff [1]. A catalytic formation of cyanoforma-
mide as an intermediate in the synthesis of oxamide
from HCN was demonstrated by Riemenschneider [2].
This process was suggested for cleaning exhaust gases
from chemical production plants. The structure of the
molecule in the crystalline phase has been determined
using X-ray crystallography [3] and in the gas phase
using electron diffraction [4]. The infrared spectrum of
the molecule in the solid phase and the Raman spectrum
of both the powdered solid and an aqueous solution
were reported in 1975 [5]. The planarity of the molecule,
demonstrated by the experimental results, is also pre-
dicted in ab initio calculations [4,6]. Measurements of
the microwave spectrum of the van der Waals complex
2. Experimental aspects
Cyanoformamide was prepared according to the
description in the patent of Yokayama et al. [10]. 4.3 g
cyanogen (NCCN) were destilled into a stirred mixture
of 0.5 g K Cr O in 20 g water and kept at 20 ꢁC for
NCCN ꢀ ꢀ ꢀ H O, a complex that can be considered to
2
be a precursor of cyanoformamide, were reported in
2
2
7
1
h. Unreacted cyanogen and water were destilled off in
1
vacuum and the remaining crystalline cyanoformamide,
Present address. Aakjars Alle 24b, DK-2860 Søborg, Denmark.
E-mail address: jjchr@tdcadsl.dk.
which melts at 58–60 ꢁC [11], was purified by sublimation
0
022-2852/$ - see front matter ꢀ 2005 Elsevier Inc. All rights reserved.
doi:10.1016/j.jms.2005.01.004

132
J.J. Christiansen / Journal of Molecular Spectroscopy 231 (2005) 131–136
in vacuum twice. Although slightly unstable, the product
could be stored at room temperature and used for inves-
tigation over several years.
large. In spite of the broad line profiles, it was possible
to measure the Stark shifts and determine the magnitude
of both components.
Slow decomposition of the sample yields, among other
products, HCN and HNCO. Thus, the sample tube accu-
mulated, after some time, several volatile substances,
which had to be pumped away before the sample could
be let into the microwave cell. Due to the low vapor pres-
sure of cyanoformamide at room temperature, 5–15 min
were allowed for filling the Stark cell. Heating with a
warm water bath was avoided, as this would have caused
the substance to sublime into the waveguide, where it
could only have been pumped out with difficulty.
The spectrometer was a Hewlett–Packard MRR spec-
trometer, model 8460 A operating in the range 18–
Furthermore, there are a large number of transitions
in excited vibrational states. Some of these overlap with
ground state lines. The combination of the hyperfine
splitting and the presence of excited state lines leads to
numerous blended and distorted lines that could be as-
signed but could not be included in the fit.
4. Quadrupole hyperfine structure
Roughly one quarter of the observed rotational lines
are split as much as 4 MHz by partially resolved electric
nuclear quadrupole hyperfine structure. The hyperfine
patterns were simulated using the program SIM2Q de-
scribed in [13] and originally written by G u¨ nther [14].
This program uses the principal values of the coupling
matrix, vaa = eQqaa (a = a,b,c), to calculate the hyper-
fine components of transitions of molecules with two
contributing nuclei. Since the spectra were not available
in digital form, and the hyperfine structure was not fully
resolved, the analysis of the nuclear hyperfine coupling
4
0 GHz and using Stark modulation in the Giessen spec-
troscopy laboratory. The spectra were recorded between
981 and 1986 and were therefore not digitized, but reg-
istered on a chart recorder.
1
3
. Spectrum and rotational assignment
Assuming a planar structure, both a- and b-type tran-
sitions could be expected. Since the structure of the mol-
ecule was known from electron diffraction and crystal
X-ray diffraction studies, estimates of the rotational
constants could be made. Another helpful model was
the study by Little and Gerry of the the microwave spec-
constants vaa and vbb ꢁ v for each nitrogen nucleus
cc
was carried out graphically. The observed hyperfine
structure was compared to simulated spectra which
were optimized iteratively. The graphical output of
the program was compared with the observed spectra
for various values over a reasonable range of hyper-
fine parameter values. This range was narrowed down
in successive steps to yield an unambiguous set of
constants.
trum of propiolamide, HCCCONH [12]. Substituting
2
HC with N, with a suitable NC internuclear distance,
gave a useful estimated structure.
A survey spectrum from 18 to 26.5 GHz was recorded
and compared with a spectrum calculated for the as-
sumed structure. A series of strong b-type Q-branch
lines could immediately be assigned. The intensity of
the observed rotational lines indicated that both the a
and b dipole moment components must be large. By
alternately fitting and assigning lines, successively more
lines of different series could be assigned and included in
the fit. A total of 41 a-type and 85 b-type clearly resolved
lines were finally assigned.
A complication, but a further source of information,
was the quadrupole hyperfine structure. Due to the cou-
pling of the quadrupole moments of two nitrogen nuclei
with the rotation, many transitions in the spectrum are
split into a large number of components, resulting in
broad line profiles, some of which show as many as five
broad components. To determine reliable line centers
for the rotational transitions, it was necessary to simu-
late the patterns of all the line profiles due to the hyper-
fine splitting. The line centers determined with the help
of these predicted patterns could then be used in a fit
to an effective rotational Hamiltonian.
One nitrogen atom is in a cyanide group, and the
other in an amide group. From the literature, it is clear
that the quadrupole coupling constants for these two
environments differ significantly, and it was therefore
possible to use this information in approaching the anal-
ysis of the hyperfine splitting.
Not all lines show hyperfine structure, and the ob-
served structure may be due to one or both nitrogen nu-
clei. Using simulations based on variations around
assumed starting values of the coupling constants, taken
from the literature, the observed lines were screened, by
running simulations with various values of the hyperfine
constants, for their degree of dependence on the cou-
pling constants of one or the other nitrogen nucleus.
This initial screening revealed that the pattern of the
b-type Q-branch transitions is determined mostly by
the amide nitrogen. The pattern of low-J b-type P-
and R-branch lines, on the other hand, was most sensi-
tive to the coupling due to the cyanide nitrogen. The
coupling constants were then refined to approximate
the observed patterns. After making use of transitions
that exhibit simple patterns to restrict the range of
parameters, the transitions showing more complicated
The intensity of the lines showed that both the a and
b components of the electric dipole moment must be

J.J. Christiansen / Journal of Molecular Spectroscopy 231 (2005) 131–136
133
patterns could then be used. A small set of lines was se-
lected, for which both the quality of the spectra and the
sensitivity to the parameters was favorable. This latter
criterion means that the number of identifiable compo-
nent peaks, their relative intensity, and the splitting be-
tween them could be clearly related to one or the other
cy
of the four parameters to be determined, v and
aa
cy
bb
cy
am
am
am
v ꢁ v , and v and vbb ꢁ v , where the superscript
cc
aa
cc
indicates the cyanide or amide group.
The simplest model system for starting values for the
cyanide nitrogen nucleus was cyanamide, with a cyanide
group directly on the a principle axis [15]. The structure
of cyanoformamide gives a tilt of the nitrogen quadrupole
environment relative to the a axis, but the projection on
the a axis should still predominate. Thus, the starting val-
cy
ues for the iterative procedure were v ¼ ꢁ3:30 MHz
aa
cy
bb
cy
cc
cy
bb
cy
cc
and v ꢁ v ¼ 0:0 MHz. The v ꢁ v value for cyana-
mide is actually nonzero, but it seemed reasonable to start
with a symmetric quadrupole tensor.
Starting values for the amide nitrogen were taken
from propiolamide [12], which differs from cyanoforma-
mide only by having an acetylene group in place of the
cyanide group. These values were thus taken as
am
aa
am
am
v
¼ 1:85 MHz and v ꢁ v ¼ 5:79 MHz.
bb
cc
The transitions selected on the basis of the initial
am
am
cc
screening for determining the value of vbb ꢁ v were
61,5
‹ 60,6, 71,6 ‹ 70,7, 71,6 ‹ 71,7, 81,7 ‹ 8 , 10
‹
0,8
2,8
1
1
02,9, 102,8 ‹ 101,9, 123,9 ‹ 123,10, 143,12 ‹ 134,9, and
am
53,13 ‹ 144,10. For determining vaa the transitions
cy
aa
‹
33,0
‹ 32,1 and 33,1 ‹ 32,2 were most suitable. For v ,
similarly, the transitions 32,2 ‹ 22,1, 33,0 ‹ 32,1, 33,1
32,2, 43,1 ‹ 42,2, and 43,2 ‹ 42,3 could be used. The two
lines 33 ‹ 3 and 33,1 ‹ 32,2 were used to determine
,0
am
2,1
cy
aa
Fig. 1. Comparison of observed (upper) and calculated (lower)
hyperfine structure in the line profile of the 33,0 ‹ 32,1 rotational
transition of cyanoformamide. The vertical line of zeros in the
simulation indicates the unsplit line position.
both vaa and v , but different features in the patterns were
used, so that the correlation between the two parameters
is actually not large. The only transition that seems to
cy
cy
show a significant dependence on vbb ꢁ v is 42,3 ‹ 41,4
.
cc
However, even this dependence is weak, so that this
parameter is not well determined. The iterative compari-
son of the observed lines with simulations of these sets
of lines, calculated for a suitable range of parameters,
led to a consistent set of hyperfine coupling constants.
Fig. 1 shows a comparison of the observed and calculated
line profiles for the 33,0 ‹ 32,1 transition. The values of the
parameters determined, together with their estimated er-
rors, are given in Table 1.
The values of the coupling constants are of interest in
their own right, but their major importance in this study
is their role in determining accurate hypothetical unsplit
rotational frequencies so that a rotational analysis could
be carried out.
5. Analysis of the rotational frequencies
For comparison, this table also includes the corre-
sponding values obtained in the principle axis system
of the molecules for propiolamide [12], cyanamide [15],
acetylcyanide [16], and formamide [17]. The values for
acetylcyanide are seen with hindsight to have presented
a better starting point than cyanamide for the refine-
With the nuclear hyperfine coupling constants given
in Table 1, the effect of hyperfine splitting on all of the
observed lines could be simulated, and the deviation be-
tween the observed maximum in the pattern for each
line and the position of the hypothetical unsplit line
could be precisely determined. Omitting transitions that
were badly overlapped, the center frequencies of 135
transitions (both components of nine nearly degenerate
asymmetry doublets are included in the fit individually
with half the weight of the other lines) corrected in this
cy
cy
cy
ment of both v and v ꢁ v . The agreement found be-
aa
bb
bb
tween the values determined for cyanoformamide and
those of acetylcyanide serves to confirm the correctness
of the present analysis.

134
J.J. Christiansen / Journal of Molecular Spectroscopy 231 (2005) 131–136
Table 1
Quadrupole coupling constants (in MHz) of cyanoformamide and of related molecules in the ground state
HCC-C(O)-NH NC-NH NC-C(O)-CH H-C(O)-NH
2
2
3
2
NC-C(O)-NH
2
propiolamide [12]
cyanamide [15]
acetylcyanide [16]
formamide [17]
cyanoformamide (this work)
cy
a
v
v
v
v
—
—
—
—
ꢁ3.30(2)
ꢁ4.34(7)
ꢁ0.28
—
—
—
—
ꢁ4.56(15)
ꢁ0.20(30)
2.18
aa
cy
bb
cy
bb
cy
cc
cy
cc
ꢁ v
2.42
2.86(2)
0.44
2.03(31)
2.31(31)
2.38
am
aa
am
bb
am
bb
am
cc
a
b
v
v
v
v
1.85(14)
5.79(7)
1.97
ꢁ3.82
3.05(2)
6.75
1.85(2)
ꢁ4.90
—
—
—
—
1.87(2)
5.83
1.63(10)
5.95(10)
2.16
ꢁ3.79
am
a
b
ꢁ vcc
b
1.98(3)
ꢁ3.85(2)
a
cy
bb
cy
cc
These values were used as starting values, and in addition the starting value of v ꢁ v was taken as zero.
b
The a and b axis labels are switched for a valid comparison with cyanoformamide.
Table 2
constants [19] was invoked to determine / , as indi-
JK
Rotational and centrifugal distortion constants obtained from a fit of
hypothetical unsplit positions of microwave rotational transitions of
cyanoformamide to WatsonÕs A-reduced I effective Hamiltonian
cated in Table 2. Omission of /JK led to a significantly
worse fit, with a standard deviation of 0.054 MHz. More
information concerning the centrifugal distortion of the
molecule will be determined from millimeter wave mea-
surements now in progress [8].
r
a
A
B
C
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
11328.38839(197) MHz
4280.667573(649) MHz
3103.006735(510) MHz
0.70903(133) kHz
D
D
D
J
JK
K
20.75683(810) kHz
ꢁ10.61323(688) kHz
0.214985(116) kHz
12.3601(114) kHz
6. Determination of the electric dipole moment
components
d
J
d
K
b
U
U
U
U
J
0.0
JK
KJ
K
62.855(556) mHz
ꢁ21.31(866) mHz
ꢁ209.5(212) mHz
The b-type lines are roughly twice as strong as the a-
type lines, so that l must be significantly larger than l .
b
a
Most of the lines thus have large Stark shifts, and well-
separated Stark lobes, but the Stark lobes are broadened
by hyperfine splitting as well as by the usual inhomoge-
neity of the electric field at high field strengths. As the
density of lines in the spectrum is fairly high, some Stark
patterns are also overlapped by other lines. It was there-
fore decided not to try to make precision measurements
of individual Stark lobes, since most are not well de-
fined. The dipole moment components were obtained
from measurements of a carefully run survey spectrum
of the region 26.5–40 GHz recorded with a Stark voltage
giving 2.16 kV/cm. The Hewlett–Packard Stark cell was
calibrated by making Stark shift measurements of OCS.
However, the accuracy of the Stark shift measurements
of cyanoformamide were limited by the width of the
lines. Suitable Stark lobes of cyanoformamide varied
in width from 2 to 8 MHz, and the accuracy of the mea-
surements taken from the chart paper (1 MHz per
0.8 mm) can thus be given as 1 MHz, in shifts between
10 and 115 MHz.
b
/
/
/
J
0.0
c
JK
K
30.0 mHz
b
0.0
Number of transitions 135; number of data 126. Standard deviation
0
.042 MHz.
a
Standard errors, corresponding to recommendations in [20], in
units of last digit.
b
Fixed at zero.
c
Determined from the planarity relation for the sextic centrifugal
distortion.
way could be used for the rotational analysis and are
listed in the Supplementary Material as Table S1. The
a-type transitions include Q-branch transitions involv-
ing levels up to J = 61 and K = 16, and R-branch tran-
sitions up to J = 5 and K = 3. The b-type transitions
extend to J = 48 and K = 22 in the P- and R-branches,
and J = 8 in the Q branch.
The effective Hamiltonian used is a semirigid rotor
Watson Hamiltonian in the I representation and A-re-
a
a
a
r
0
r
duction as formulated in the program AWAT [18].
The resulting constants are given in Table 2. One of
the primary results of the analysis is that the molecule
An iterative procedure was used to determine the val-
ues of the dipole moment components. Calculations
were carried out at Chemical Laboratory V, University
of Copenhagen with the program STARK. For a given
field strength and values of the dipole moment compo-
nents, the program calculates the quadratic Stark shift,
the correction for the linear effect due to near degenera-
cies, and in addition the coefficients of the quadratic
2
˚
is indeed planar: The inertial defect is 0.195 amu A ,
indicating clearly that the NH group is quite rigid.
2
Since this is the case, and the seven independent sextic
centrifugal distortion constants were not well-defined
when all were adjusted, the planarity condition for these

J.J. Christiansen / Journal of Molecular Spectroscopy 231 (2005) 131–136
135
Fig. 2. Region of the 122,9 ‹ 123,10 transition of cyanoformamide measured with Stark modulation at 2.16 kV/cm, showing Stark lobes (extending
upwards) to higher frequency.
contribution for each M value, that is for each individ-
ual Stark lobe. An initial calculation was made with la
To derive a correction for the values of the dipole
moment components used in the previous calculation,
the deviations between the shifts and their predictions
for each group of lines were added, as were the linear
equations describing the quadratic contributions of each
component. Depending on how the groups of lines were
defined, this led to two, three or four linear equations.
Any set of two of these could easily be solved to yield
and l set equal to 1 Debye, to obtain values for the
b
coefficients in the quadratic Stark effect for each lobe
of candidate lines. After a few easily identified Stark
shifts had been measured, these coefficients could then
be used to determine starting values for l and l from
a
b
the observed shifts. A new calculation, based on these
values and the correct electric field strength of
2
2
correction values for l and l . The calculation and
a
b
2
.16 kV/cm, was then used to search for and select those
the correction procedure were reiterated a few times,
and a consistent solution was obtained for the values
l = 2.25(10) D and l = 3.43(10) D. Because of the un-
even quality of the Stark shift data, and the limited rig-
orousness of the procedure, we have given rather large
error margins, but the robustness of the solution to dif-
ferent selections of lines indicates that these values are
very well defined.
Stark lobes which would serve as data for refining the
values. After one or two further iterations, 22 Stark
lobes of seven rotational transitions were selected. The
lobes of 113,8 ‹ 112,9, for M = 4, 7, and 8, and
a
b
1
23,9 ‹ 122,10, M = 6 through 12, all exhibit a similar
relative dependence on l and l , with l contributing
a
b
b
about six times as much as l . The region of the transi-
a
tion 123,9 ‹ 122,10 is shown in Fig. 2. The lobes
1
03,7 ‹ 102,8, M = 5, 6 and 7 and 144,10 ‹ 143,11,
M = 5, 7, 8, and 9, again each exhibit a similar relative
dependence on l and l , with l contributing in this
7. Discussion
a
b
b
case about three times as much as l . The lobes
The present work has confirmed unequivocally the
planarity of cyanoformamide, and has led to a determi-
nation of the quadrupole coupling parameters and di-
pole moment components. The probability that this
molecule might be detected in the interstellar medium
is significant, considering the abundances of related spe-
cies with the same functional groups, and the recent in-
creased detections of larger molecules. The present study
a
52,4
‹ 42,3, M = 1 and 2 show contributions of l that
b
are, respectively, only 1.5 and 0.35 as large as those of
la, and with opposite sign. Stark lobes from two further
transitions, M = 1 from the transition 50,5 ‹ 40,4, and
M = 0 and 1 from the transition 40,4 ‹ 30,3, show con-
tributions of l and l that differ from the above and
b
a
were also used.

136
J.J. Christiansen / Journal of Molecular Spectroscopy 231 (2005) 131–136
provides the parameters needed for an interstellar search
in the microwave region. Furthermore, it has facilitated
a current extensive study of the rotational spectrum in
the millimeter and submillimeter wave range, which will
ultimately be essential for interstellar measurements [8].
Transitions from three excited vibrational states were
also identified in the spectrum, and will be the subject of
a separate publication [9].
References
[
[
[
1] N. Beketoff, Chem. Ber. 3 (1870) 872–873.
2] W. Riemenschneider, Chemtech (1976) 658–661.
3] U. Dr u¨ ck, W. Becker, G. Becker, Z. Kristallogr 167 (1984) 131–
1
34.
[
[
4] S. Saebø, B. Klewe, S. Samdal, Chem. Phys. Lett. 97 (1983) 499–
502.
5] H.O. Desseyn, F.K. Vansant, B.J. Van Der Veken, Spectrochim.
Acta 31A (1975) 625–633.
[
[
6] K.-T. Lim, M.M. Francl, J. Phys. Chem. 91 (1987) 2716–2721.
7] S. Lee, I.I. Suni, W. Klemperer, J. Chem. Phys. 96 (1992) 5577–
Acknowledgments
5
584.
[
[
8] M. Winnewisser, I.R. Medvedev, F.C. De Lucia, E. Herbst, J.
Koput, K.V.L.N. Sastry, R.A.H. Butler, Astrophys. J. Suppl., in
press.
9] M. Winnewisser, I. Medvedev, F.C. De Lucia, E. Herbst, J.J.
Christiansen, Talk RC09, 59th Ohio State University Interna-
tional Symposium on Molecular Spectroscopy, June (2004) 21–
The author is grateful for the support and encourage-
ment of Manfred Winnewisser for the opportunity to
carry out this work in the Giessen spectroscopy labora-
tory, and to Brenda P. Winnewisser for discussions of
the hyperfine structure and for help in completing the
manuscript. The program AWAT, written by K.M.T.
Yamada, was made available by the Giessen laboratory.
The use of the computer programs ROTFIT and
STARK, made available by G.O. Sørensen, and facili-
ties at Chemical Laboratory V, University of Copenha-
gen is gratefully acknowledged. The work in Giessen
was supported in part by the Deutsche Forschungs-
gemeinschaft and the Fonds der chemischen Industrie
through the grants of Manfred Winnewisser.
2
5.
[10] S. Yokayama, K. Tanaka, S. Furuhashi, Japan. Kokai Patent No.
74,133,326; Chem. Abstr. 82: P139 436h, (1975).
11] R.P. Welcher, M.E. Castellion, V.P. Wystrach, J. Am. Chem.
[
Soc. 81 (1959) 2541–2547.
[
[
12] G.B. Little, M.C.L. Gerry, J. Mol. Spectrosc. 71 (1978) 321–342.
13] M. Winnewisser, B.P. Winnewisser, C. Wentrup, J. Mol. Spec-
trosc. 105 (1984) 193–205.
[14] H. G u¨ nther, Dissertation, Universit a¨ t T u¨ bingen (1975).
15] J.N. Macdonald, D. Taylor, J.K. Tyler, J. Sheridan, J. Mol.
[
Spectrosc. 26 (1968) 285–293.
[
[
16] F. Scappini, H. Dreizler, Z. Naturforsch. 31a (1976) 840–846.
17] W.H. Kirchhoff, D.R. Johnson, J. Mol. Spectrosc. 45 (1973) 159–
1
65.
Appendix A. Supplementary data
[18] K.M.T. Yamada, S. Klee, J. Mol. Spectrosc. 166 (1994) 395–405.
[
[
19] W. Gordy, R.L. Cook, Microwave Molecular Spectra, third ed.,
John Wiley, New York, 1984, p. 368.
20] D.L. Albritton, A.L. Schmeltekopf, R.N. Zare, An Introduction
to the Least-Squares Fitting of Spectroscopic Data, in: K.
Narahari Rao (Ed.), Molecular Spectroscopy: Modern Research,
Vol 2, Academic Press, New York, 1976.
Supplementary data for this article are available on
ScienceDirect (www.sciencedirect.com) and as part of
the Ohio State University Molecular Spectroscopy Ar-
chives (http://msa.lib.ohio-state.edu/jmsa_hp.htm).