Journal of Molecular Spectroscopy 204, 235–261 (2000)

doi:10.1006/jmsp.2000.8226, available online at http://www.idealibrary.com on

Jet-Cooled Optical Spectroscopy of FeN

between 16 300 and 21 600 cm ⴚ1
Kosuke Aiuchi and Kazuhiko Shibuya
Department of Chemistry, Graduate School of Science and Engineering, Tokyo Institute of Technology,
2-12-1 Ohokayama, Meguro-ku, Tokyo 152-8551, Japan

Received June 8, 2000; in revised form August 23, 2000

We report the first gas-phase spectroscopic study of iron mononitride (FeN). FeN molecules were generated by the reaction
of laser-ablated Fe atoms with NH 3 under supersonic jet-cooled conditions. Laser-induced fluorescence excitation spectra were
measured between 16 300 and 21 600 cm ⫺1, and rotational analyses have been performed for about 25 vibronic bands. The ⍀
value of the ground state has been determined to be 25, which could be the spin– orbit component of 2⌬ (. . . 1␦ 39␴ 2) or 4⌸
(. . . 1␦ 39␴ 14␲ 1). The excited states in the visible absorption region show very complicated rovibronic structures due to heavy
perturbations. Excited state lifetimes, 56Fe 14N/ 56Fe 15N isotope shifts, and dispersed fluorescence spectra were measured for the
majority of the bands, which were classified into the five ⍀ ⫽ 25–X⍀ ⫽ 25 band systems. © 2000 Academic Press
Key Words: FeN; laser-induced fluorescence; dispersed fluorescence.

1. INTRODUCTION the averaged quadratic coupled cluster (AQCC) method (30).

In recent years, there has been considerable interest in the
study of transition-metal-containing molecules because of their
importance in astrophysics (1– 4) and catalytic processes (5).
Diatomic-transition metal nitrides are the simplest model sys-
tems in which the transition-metal–nitrogen bond is present,
and the spectroscopic and theoretical studies of these mole-
cules provide important information about the bonding be-
tween a metal atom and a nitrogen atom. Although some
transition-metal hydrides and oxides have been detected in
stellar atmospheres, transition-metal nitrides have not been
observed. Because nitrogen has a large cosmic abundance and
many transition-metal-containing molecules have been identi-
fied in spectra from cool M- and S-type stars, there is a
possibility that transition metal nitrides will be found. A num-
ber of transition metal nitrides have recently been studied. For
the 3d transition metal family, ScN (6), TiN (7–13), VN
(14 –17), and CrN (17–19) have been studied in the gas phase.
Theoretical calculations for ScN (20 –21), TiN (22–26), VN
(26, 27), CrN (26, 28), FeN (28 –30), NiN (29), and CuN (31)
have also been reported. For the VIIIB transition metal family,
RuN (32) and OsN (33) have been studied both experimentally
and theoretically.
For FeN, Blomberg and Siegbahn calculated several low-
lying electronic states ( 2⌬, 2⌸, 2⌺ ⫹, 4⌬, 4⌸, and 4⌺ ⫹) with the
complete active-space self-consistent field (CASSCF) method,
and three of these ( 2⌬, 2⌺ ⫹, and 4⌸) with the internally con-
tracted averaged coupled-pair function (IC-ACPF) method
(28). They predicted that the 2⌬ state is the ground state with
the 4⌸ state only 800 cm ⫺1 higher in energy. More recently,
Fiedler and Iwata calculated 24 low-lying electronic states with
235
They reported that the 2⌬ state is the lowest in energy and the
three low-lying 6⌺ ⫹, 4⌸, and 4⌽ states were calculated to be
located 180, 1050, and 1750 cm ⫺1 higher in energy, respec-
tively. They concluded that these four states are all good
candidates for the ground electronic state of FeN because the
energy differences between these states might be smaller than
the accuracy of the calculation method.
In this study, laser-induced fluorescence (LIF) excitation
spectra of jet-cooled 56Fe 14N and 56Fe 15N have been recorded in
the region between 16 300 and 21 600 cm ⫺1. About 25 bands
have been assigned to FeN, and rotational analyses have been
carried out. The ground state of FeN has been shown to have
the ⍀ value of 25, for which the most probable candidate is the
spin– orbit component of a 2 ⌬ i state. Excited state lifetimes and
dispersed fluorescence (DF) spectra have also been measured
for the majority of the observed bands. To our knowledge, the
work reported here is the first spectroscopic study of FeN in the
gas phase.
2. EXPERIMENTAL DETAILS
FeN molecules were generated by the reaction of laser-
ablated Fe atoms with NH 3 under supersonic jet-cooled con-
ditions, and their spectra were recorded by LIF experiment.
A pulsed valve (General Valve, P/N 9-279-900) with an
0.8-mm-diameter orifice was used with a rotating iron rod
(5-mm diameter) mounted near the opening of the valve. A
NH 3/He gas mixture (3% ammonia in 2.5 atm helium) was
expanded through an attached channel (24 mm long and 2 mm
in diameter) into a vacuum chamber. The background pres-
sures were 2 ⫻ 10 ⫺4 Pa with the valve closed and 1 ⫻ 10 ⫺2 Pa
0022-2852/00 $35.00
Copyright © 2000 by Academic Press
All rights of reproduction in any form reserved.
236
FIG. 1. A schematic representation of the vibronic bands of 56Fe 14N observed in the 16 000 –21 800 cm ⫺1 region. Vibrational assignment of the band systems
(I) ⬃ (V)⍀ ⫽ 25–X⍀ ⫽ 25 are indicated.
during the valve operation. The output of a KrF excimer laser
(Lambda Physik, COMPex 102; 248 nm) was used as a vapor-
ization light source.
To measure the LIF excitation spectra of FeN, a XeCl
excimer laser (Lambda Physik, COMPex 102; 308 nm)
pumped dye laser (Lambda Physik, SCANMATE) was used to
excite jet-cooled FeN. An étalon was placed in the cavity of the
dye laser to narrow the linewidth to 0.04 cm ⫺1, and the wave-
number calibration was accomplished using the Te 2 atlas (34)
or the I 2 atlas (35). The fluorescence was detected with a
photomultiplier tube (Hamamatsu, R928) and the PMT signal
was sent to a boxcar integrator (Stanford Research, SR 250),
where the signals were averaged for 10 sample shots. The
measured spectra were mainly due to the naturally abundant
56
FeN (91.7%), and the contributions of the minor species such
as 54FeN (5.8%) were negligible. To measure the spectra of
56
Fe 15N, 15NH 3 (Nippon Sanso, 99%) was used. For 56Fe 15N,
only five bands were recorded at a resolution of 0.04 cm ⫺1 to
determine the accurate rotational constant for the ground state,
and low-resolution (0.2 cm ⫺1) spectra were recorded for the
majority of the bands.
To measure the DF spectra, the excitation laser was fixed at
a region of strong fluorescence within the band of interest, and
the grating of a monochromator (Nikon, G250) was scanned.
The slitwidths of 0.5–1.2 mm were used; the corresponding
resolution was 60 –120 cm ⫺1. The monochromator was cali-
brated by the excitation laser wavelength. The LIF decay
curves were also recorded by averaging the signals for 256
laser shots. Each curve was fitted to a single exponential decay
model using a nonlinear least-squares algorithm, and the life-
time of the excited state was extracted.
Copyright © 2000 by Academic Press
AIUCHI AND SHIBUYA
3. APPEARANCE OF LIF EXCITATION SPECTRA
The LIF excitation spectra were recorded in the region
between 15 300 and 22 830 cm ⫺1. In addition to the well-
known visible bands of NH 2, a number of vibronic bands not
previously reported to our knowledge have been observed. The
lower state rotational constants of these bands are almost the
same, indicating that the state from which these transitions
originate is identical for each band. The carrier of these bands
is unambiguously assigned as FeN from the rotational constant,
the half-integer angular momentum J, and the experiment
using 15NH 3 and 14ND 3. The B 0 ( 56 Fe 15N)/B 0 ( 56 Fe 14N) ratio for
the lower state is determined to be 0.9470, which is quite close
to the mass-spectrometric ␳ 2 value of 0.9468. The spectra taken
using ND 3 are identical to those taken using NH 3, showing that
no hydrogen is present in the species of the present interest.
The B 0 value of 0.602793(17) cm ⫺1 determined for 56Fe 14N is
similar to those of the other 3d transition-metal nitrides which
have been published previously (6, 13, 14, 18).
Figure 1 shows a schematic representation of the observed
spectrum of 56Fe 14N. Because it is synthesized from the spectra
recorded in a lot of separate experiments, the relative intensi-
ties compared over wide wavelength ranges are considered to
be rough. The lowest transition energy band appeared at 16 342
cm ⫺1, and no band appeared between 15 300 and 16 342 cm ⫺1.
Only the two bands observed at 21 514 and 21 552 cm ⫺1 are
⍀ ⫽ 23–⍀ ⫽ 25 transitions and the other 23 bands are ⍀ ⫽ 25–⍀ ⫽
2 transitions. The ⍀ quantum numbers for the upper and the
5
lower states were clearly determined from the first J values of
the P, Q, R lines. Observation of the ⌳-doublings indicates
that these states are doubly degenerate. It is reasonable to
consider that the lower ⍀ ⫽ 25 state is the ground electronic
 JET-COOLED OPTICAL SPECTROSCOPY OF FeN 237

TABLE 1
Spectroscopic Constants Determined for the Rotationally Resolved Bands of 56Fe 14N a

a
The rotational constants of the ground state are B ⬙eff ⫽ 0.602793(17) cm ⫺1 and D ⬙eff ⫽ 1.211(41) ⫻ 10 ⫺6 cm ⫺1. In
all cases, quoted error limits represent 1␴ in the least-squares fit of the data.
b
Isotope shift determined as ␯ 0( 56Fe 14N) ⫺ ␯ 0( 56Fe 15N).
c
Determined from a deperturbation analysis using a homogeneous perturbation model. For the 20560 cm ⫺1 pair of
levels, H 12 is determined to be 0.6931(44) cm ⫺1. For the 20670 cm ⫺1 pair of levels, H 12 is determined to be 0.6817(27)
cm ⫺1. For the 21380 cm ⫺1 pair of levels, H 12 is determined to be 1.3710(95) cm ⫺1. The ⍀ value of the perturber is
ambiguous for the 20670 cm ⫺1 pair of levels.

state of FeN because our experiments were conducted under
jet-cooled conditions.
All the bands are red-degraded due to a significant geometry
change upon electronic excitation. The vibronic structure is
very complicated due to a multitude of perturbations, making
the identification of band systems difficult. Vibrational assign-
ment of the band systems from (I)–X to (V)–X indicated in Fig.
1 is based on the excited state lifetimes, the isotope shifts, and
the Franck–Condon patterns in the DF spectra, which will be
discussed later. Although we recorded the spectra up to 22 830
cm ⫺1, no vibronic bands of FeN were observed above 21 600
cm ⫺1. This may be due to the fact that some predissociation
channel is open in this region.
4. ROTATIONAL ANALYSES
Rotational analyses of 56Fe 14N and 56Fe 15N are summarized
in Tables 1 and 2, and the line positions of 56Fe 14N and 56Fe 15N
are provided in Appendixes 1 and 2, respectively. Since the
rotational structures of the spectra are very complicated due to
the heavy rotational perturbations in the upper states, the
rotational constants of the ground state were determined using
all available combination differences. The energy level expres-
sion for the ground state is
F共 J兲 ⫽ B eff J共 J ⫹ 1兲 ⫺ D eff 关 J共 J ⫹ 1兲兴 2 . [1]
The maximum J⬙ value is 20.5, and no evidence was found for
⌳-doubling. The derived rotational constants of the lower
states are B ⬙eff( 56Fe 14N) ⫽ 0.602793(17) cm ⫺1, D ⬙eff( 56Fe 14N) ⫽
1.211(41) ⫻ 10 ⫺6 cm ⫺1, B ⬙eff( 56Fe 15N) ⫽ 0.570847(40) cm ⫺1,
and D ⬙eff( 56Fe 15N) ⫽ 1.102(85) ⫻ 10 ⫺6 cm ⫺1.
Upper state rotational constants and band origins were then
determined band by band by least-squares fit, where rotational
constants of the lower state were fixed. Many perturbed lines
were excluded in the fit. This can be seen in Appendixes 1 and
2, in which O ⫺ C values are listed only for the lines employed

Copyright © 2000 by Academic Press

238 AIUCHI AND SHIBUYA

TABLE 2
Spectroscopic Constants Determined for the Rotationally Resolved Bands of 56Fe 15N a

a
The rotational constants of the ground states are B ⬙eff ⫽ 0.570847(40) cm ⫺1 and D ⬙eff ⫽ 1.102(85) ⫻
10 cm ⫺1. In all cases, quoted error limits represent 1␴ in the least-squares fit of the data.
⫺6

b
Determined from a deperturbation analysis using a heterogeneous perturbation model where ⌬⍀ ⫽ ⫺1
is assumed (see text). The coefficient k is determined to be 0.2007(19) cm ⫺1.
c
Determined from a deperturbation analysis using a 3 ⫻ 3 matrix model. The ⍀ assignment of the
perturber B is ambiguous. The off-diagonal elements are determined to be H main,A ⫽ 3.890(13) cm ⫺1,
H main,B ⫽ 1.904(28) cm ⫺1 in the fit. H A,B is fixed at 0 cm ⫺1.
d
Determined from a deperturbation analysis using a homogeneous perturbation model. For the 19 880
cm ⫺1 pair of levels, H 12 is determined to be 0.878(26) cm ⫺1. For the 21380 cm ⫺1 pair of levels, H 12 is
determined to be 0.7381(22) cm ⫺1. In both cases the ⍀ values of the perturbers are ambiguous.

for the fit. The energy-level expression for the upper states is
the same as that for the lower state, but D eff was set to zero for
the majority of the bands since the perturbations in the upper
states limit the usable range of J values. Bands for which D eff
was included for the fit are the 16 342 cm ⫺1 band of 56Fe 14N
and the 16 344 and 16 683 cm ⫺1 bands of 56Fe 15N.
For seven of the observed bands, deperturbation analyses
were carried out to obtain better B⬘eff values. A 2 ⫻ 2 matrix
model was used for the 20 560, 20 670, and 21 380 bands of
56
Fe 14N and the 17 700, 19 880, and 21 380 cm ⫺1 bands of
56
Fe 15N. A 3 ⫻ 3 matrix model was used for the 19 610 cm ⫺1
band of 56Fe 15N. The upper state rotational structures are
shown in Fig. 2 for 56Fe 14N and Fig. 3 for 56Fe 15N.
The spectrum of the 20 560 cm ⫺1 band is shown in Fig. 4,
and the upper state rotational structure is shown in Fig. 2a. This
is the simplest example because other perturbers do not exist
around the crossing point ( J ⫽ 3.5–4.5). Small perturbations
are seen at J ⫽ 11.5, and other perturbations also exist at J ⱖ
15.5 (not displayed in Fig. 2a), so that these lines are excluded
from the fit. Because the lowest J line of the perturber was not
observed in this band, it is not possible to determine the J
dependence of the off-diagonal element clearly. The J-depen-
dent off-diagonal element is expressed as (38)
具J⍀兩H兩J⍀ ⫾ 1典 ⫽ k 冑J共 J ⫹ 1兲 ⫺ ⍀共⍀ ⫾ 1兲.
[2]
Assuming that the essential intensity of the transition to the
perturbing state is zero, we can deduce the intensity ratio of
main line to extra line at a certain J value from the eigenvectors
of the Hamiltonian matrices for the two possibilities. In this
case, where the extra lines are observed at J ⫽ 3.5 and 4.5, the

Copyright © 2000 by Academic Press

 JET-COOLED OPTICAL SPECTROSCOPY OF FeN 239

For the 17 700 cm ⫺1 band of 56Fe 15N, shown in Fig. 3a, a

J-dependent model was found to be more appropriate. How-
ever, we could not decide whether ⌬⍀ ⫽ ⫺1 or ⫹1 is appro-
priate, since the lowest J lines of the perturber were not
observed and the two choices give almost the same intensity
pattern.
For the 19 610 cm ⫺1 band of 56Fe 15N, shown in Fig. 3b, a
3 ⫻ 3 matrix model was used because sufficient extra lines
were observed around the first and second crossing points ( J ⫽
7.5 and 11.5). The first perturber has been found to have an ⍀
value of 52 because the lowest J lines were observed. We could
not determine the ⍀ value of the second perturber, and a
J-independent matrix element was used provisionally for the
interaction between the main levels and the second perturber.
The interaction between the two perturbers was neglected. In
addition, small perturbations exist for the lowest energy level
around J ⫽ 5.5, and therefore the J⬘ ⫽ 4.5–6.5 levels of the
lowest energy state were excluded in the fit. J⬘ ⫽ 14.5 and
15.5 levels were also excluded in the fit because the corre-
sponding O ⫺ C values were large. This may be due to the fact
that another perturber exists at larger J⬘. For some lines which
show irregular ⌳-doublings, averages of the line frequencies
were used for the fit.

FIG. 2. Upper state level structure of the (a) 20 560 cm ⫺1, (b) 20 670
cm ⫺1, and (c) 21 380 cm ⫺1 bands of 56Fe 14N. The courses of the deperturbed
levels are marked by straight lines.

experimental and calculated intensity ratios (⫽ higher energy

line:lower energy line) are

Experiment ⌬⍀ ⫽ 0 ⌬⍀ ⫽ ⫺1 ⌬⍀ ⫽ 1

J ⫽ 3.5 24:76 22:78 12:88 9:91

J ⫽ 4.5 64:36 63:37 51:49 47:53
Apparently, ⌬⍀ ⫽ 0 agrees most with experiment, thus we use
the J-independent model in this case.
For the 20 670 cm ⫺1 band of 56Fe 14N, shown in Fig. 2b, in
addition to the first avoided crossing at J ⫽ 5.5, a lot of
irregular structures are seen at J ⱖ 7.5. Our approach has been
to truncate the branch data at J ⫽ 6.5 in the fit, and the
averages of the line frequencies have been taken for the lines
in which ⌳-doublings exist ( J ⫽ 5.5, 6.5). In this case, a
J-independent model was used, but we could not decide which
model is appropriate.
The 21 380 cm ⫺1 band of 56Fe 14N, shown in Fig. 2c, also
shows a complicated structure. The first avoided crossing ex-
ists at J ⫽ 5.5, and other irregular structures can be recognized
at J ⱖ 10.5. In this case, we truncated the branch data at J ⫽
9.5 in the fit. Analysis has shown the J-independent model to FIG. 3. Upper state level structure of the (a) 17 700 cm ⫺1, (b) 19 610
be more appropriate for this band. cm ⫺1, (c) 19 880 cm ⫺1, and (d) 21 380 cm ⫺1 bands of 56Fe 15N.

Copyright © 2000 by Academic Press

240 AIUCHI AND SHIBUYA

FIG. 4. An LIF spectrum of the ⍀ ⫽ 25–X 2 ⌬ 5/ 2 transition of 56Fe 14N observed at 20 560 cm ⫺1. The upper state is perturbed at J⬘ ⫽ 3.5–4.5. Small local
perturbation can also be seen at J⬘ ⫽ 11.5.

For the 19 880 and 21 380 cm ⫺1 band of 56Fe 15N (shown in
Figs. 3c and 3d, respectively), a J-independent model was
used, but the ⌬⍀ assignments are ambiguous in both cases. For
the 21 380 cm ⫺1 band, the branch date was truncated at J ⫽
5.5 in the fit because many perturbations existed at higher J
values.
Except for the 17 700 and 19 880 cm ⫺1 bands of 56Fe 15N,
very small B eff values ranging from 0.29 to 0.37 cm ⫺1 were
obtained for the perturbing states. These states are considered
to be high vibrational levels of lower lying electronic states. As
shown in the next section, isotope shifts do not seem to help to
make the vibrational assignment for many bands because of the
heavy perturbations. Moreover, the corresponding bands of
isotope species could not be identified for several bands. For
example, we could not observe the vibronic band of 56Fe 15N
which correspond to the 17 078 cm ⫺1 band of 56Fe 14N. This
band is overlapped by the more intense 17 087 cm ⫺1 band,
from which it merely seems to borrow its transition intensity.
Similarly, the 20 431 and 20 801 cm ⫺1 bands of 56Fe 14N for
which the corresponding bands of 56Fe 15N were not observed
seem to borrow their intensity from the 20 447 and 20 806
cm ⫺1 bands, respectively.
5. FRANCK–CONDON PATTERNS IN DF SPECTRA AND
VIBRATIONAL ASSIGNMENTS
DF spectra were recorded for the majority of observed bands
of 56Fe 14N. These spectra do not show the simple structures of
a typical diatomic molecule because the fluorescence termi-
nates not only on the ground state but also on some low-lying
electronic states. In addition to the vibrational progression of
the ground electronic state, three progressions, beginning at
545, 1407, and 1625 cm ⫺1 above the ground state, were ob-
served. The vibrational frequencies of these states are listed in
Table 3, in which three low-lying excited electronic states are
designated as a, b, and c. The assignments of the three
low-lying states are difficult because the rotational structures of
these states have not been observed. However, much informa-
tion with which the complicated excitation spectra are classi-
fied into band systems has been obtained from the DF spectra.

TABLE 3
Molecular Constants Derived for the Low-Lying Electronic States of 56Fe 14N

Copyright © 2000 by Academic Press

 JET-COOLED OPTICAL SPECTROSCOPY OF FeN 241

analysis of the (III)–(V)⍀ ⫽ 25 states because they also show

FIG. 5. DF spectra from the (I) and (II)⍀ ⫽ 25 states of 56Fe 14N. These
spectra mainly show the vibrational structure of the X⍀ ⫽ 25 state.
very irregular vibrational spacings, or only the ⌬G 1/ 2 value,
which seems not appropriate for the intrinsic ␻ e value is
obtained. Here, small changes of the vibrational constants do
not much affect the Franck–Condon patterns for these low v
values, while the change of the rotational constant (i.e., inter-
nuclear distance) is more effective. The results are shown in
Fig. 6. Since the rotational constants of the (I) and (II) states
are similar, the calculated Franck–Condon patterns resemble
each other. Although the DF spectra have not been corrected
for the monochromator and photomultiplier tube sensitivities,
the node patterns of the DF spectra can be compared with the
calculated pattern. The 17 087, 17 725, 18 345, and 18 743
cm ⫺1 bands can be assigned as the 1– 0, 2– 0, 2– 0, and 3– 0
bands, respectively. However, it is difficult to decide to which
of the systems they belong. The irregular vibrational structure
may be partly due to the perturbation between the (I) and (II)
states, while the severely perturbed rotational structures indi-
cate that other perturbers may contribute to this. The isotope
shifts of 16 and 36 cm ⫺1 for the 17 087 and 18 345 cm ⫺1 bands
are consistent with the vibrational assignment of 1– 0 and 2– 0,
respectively. The shift of 22 cm ⫺1 for the 17 725 cm ⫺1 band is
rather closer to the value of the 1– 0 band, although it is
assigned as the 2– 0 band.
The v⬘ ⫽ 0 level lifetimes of the (I) and (II) states of 56Fe 14N
are 0.43 and 0.72 ␮s, respectively. The corresponding lifetimes

The ⍀ ⫽ 52–X⍀ ⫽ 25 transitions are classified into band systems

from (I)–X to (V)–X.

(I)⍀ ⫽ 25 and (II)⍀ ⫽ 25 States

The DF spectra from the six excited states observed in the
region between 16 340 and 18 750 cm ⫺1 are shown in Fig. 5.
These spectra mainly show the vibrational structure of the
X⍀ ⫽ 52 state. The DF spectra from the excited states observed
at 16 342 and 16 682 cm ⫺1 show the typical Franck–Condon
pattern of v⬘ ⫽ 0. Moreover, very small isotope shifts of ⫺1.9
and ⫺1.3 cm ⫺1 have been obtained for the 16 342 and 16 682
cm ⫺1 bands, respectively, showing that these are evidently 0 – 0
bands. These ⍀ ⫽ 25 states are designated as (I) and (II). The DF
spectra from the other four excited states show a Franck–
Condon pattern of v⬘ ⬎ 0, and it is likely that these bands
belong to the (I)⍀ ⫽ 25–X⍀ ⫽ 5/ 2 or (II)⍀ ⫽ 25–X⍀ ⫽ 25
system. The classification of these bands is difficult because
they show very irregular vibrational spacings.
To make the vibrational assignment, Franck–Condon factors
兩具v (I)兩v X 典兩 2 and 兩具v (II)兩v X 典兩 2 were calculated using Morse curves.
Because the vibrational constants of the upper states could not
be obtained, the ␻ e values of the upper states were deduced
from that of the ground state using the empirical equation
r 2 ␻ e ⫽ const. (36), and the ␻ e x e values of the upper states
were set at 7.2 cm ⫺1, which is the average of the values of the FIG. 6. Calculated Franck–Condon factors of 56
Fe 14N. F and Œ,
X, a, b, and c states. This procedure was also used for the 兩具v (I)兩v X 典兩 2 and 兩具v (II)兩v X 典兩 2 , respectively.

Copyright © 2000 by Academic Press

242 AIUCHI AND SHIBUYA

more severe. There is evidence that the (IV)⍀ ⫽ 25 (v⬘ ⫽ 1)

state is considerably perturbed. The DF spectrum of the 20 364
cm ⫺1 band is almost the same as that of the (IV)1– 0 band. The
DF spectrum of the 20 447 cm ⫺1 band shows the mixed struc-
ture of that of the (IV)1– 0 band and the (V)0 – 0 band. These
vibronic levels are considered to be the perturbers of the
(IV)⍀ ⫽ 25 (v⬘ ⫽ 1) state. The lifetimes of the (III) and (IV)
states are 0.7–1.3 and 0.21– 0.37 ␮s, respectively.

(V)⍀ ⫽ 52 and the ⍀ ⫽ 23 States

FIG. 7. DF spectra from the (III) and (IV)⍀ ⫽ 25 states of 56Fe 14N. F,
Calculated Franck–Condon factors where a value of 1.74 Å was used for r 0 (a).
of 56Fe 15N are similar to those of 56Fe 14N. The four vibra-
tionally excited levels of the (I) or (II) states have substantially
different lifetimes ranging from 0.6 to 1.1 ␮s, and the agree-
ments between two isotopic species are not good, indicating
that many perturbations exist.
The 0 – 0 and 1– 0 bands of the (V)–X transition are most
prominent in the LIF excitation spectrum. Surprisingly, the DF
spectra from these states do not show the vibrational structure
of the X state, and only the structure of the b state can be
identified. The DF spectra and the calculated Franck–Condon
factors are shown in Fig. 8. The value of 1.66 Å for r 0 (b) was
used for the best agreement, but this value does not seem to be
accurate because the B 0 value of the (V)⍀ ⫽ 52 state is a
perturbed value. The isotope shifts for these bands do not agree
with the values expected from the vibrational assignments,
indicating that the perturbations are severe. The lifetimes of the
(V) states are 0.12– 0.18 ␮s.
The DF spectra of the 20 807 and 21 413 cm ⫺1 bands are
almost the same as those of the (V)0 – 0 and 1– 0 bands,
respectively. These bands merely seem to borrow their transi-
tion intensities from the (V)⍀ ⫽ 25–X⍀ ⫽ 52 transition.
The DF spectra from the two ⍀ ⫽ 23 states are shown in Fig.
8. These spectra show the vibrational structure of the c state,

(III)⍀ ⫽ 25 and (IV)⍀ ⫽ 25 States

Only the v⬘ ⫽ 0 and 1 levels are identified for the (III)⍀ ⫽
5
2 state, and the vibrational spacings are very irregular for the
(IV)⍀ ⫽ 25 state for which the v⬘ ⫽ 0 –2 states are identified.
The DF spectra from these states are shown in Fig. 7. These
spectra mainly show the vibrational structure of the a state.
Using the observed rotational constants and deduced vibra-
tional constants, Franck–Condon factors 兩具v (III)兩v a 典兩 2 and
兩具v (IV)兩v a 典兩 2 were calculated where the internuclear distance for
the a state was varied to reproduce the experimental patterns.
Best agreement was found using a value for r 0 (a) of 1.74 Å,
and the results are included in Fig. 7. Although the internuclear
distances of the (III) and (IV) states used for calculation were
somewhat ambiguous due to perturbations, good agreement
can be found simultaneously for both (III) and (IV) states.
Compared with the value of 1.580 Å for the X⍀ ⫽ 25 state, the
internuclear distance is significantly lengthened in the a state.
The isotope shifts for the five bands of the (III)–X and (IV)–X
transitions are not consistent with the values expected from the FIG. 8. DF spectra from the (V)⍀ ⫽ 25 states and the ⍀ ⫽ 23 states of
vibrational assignments, indicating that the perturbations are 56
Fe 14N. F, 兩具v (V)兩v b 典兩 2 , where a value of 1.66 Å was used for r 0 (b).

Copyright © 2000 by Academic Press

 JET-COOLED OPTICAL SPECTROSCOPY OF FeN
FIG. 9. The low-lying electronic states of FeN calculated by Fiedler and
Iwata (30). Expected spin– orbit splittings are included (see text).
and they are both likely to show the Franck–Condon patterns of
v⬘ ⫽ 1. Here, we could not find an r 0 (c) value for which these
Franck–Condon patterns agree well with both the calculated
patterns simultaneously. The patterns of the DF spectra of the
21 514 and 21 553 cm ⫺1 bands can be reproduced with values
of 1.62 and 1.59 Å for r 0 (c), respectively, where the B values
of these states are used instead of B 0 . However, these internu-
clear distances are of course quite ambiguous.
6. DISCUSSION
The assignments of the observed low-lying electronic states of
FeN are troublesome because the rotational structures of the a, b,
and c states have not been observed. Previous theoretical works
also predicted a complicated electronic structure. Figure 9 shows
the low-lying electronic states of FeN calculated by Fiedler and
Iwata (30), where the spin– orbit splittings estimated as below are
included. In the single configuration approximation, the following
diagonal matrix elements are expected:
1
具 2 ⌬ 5/ 2 兩H SO兩 2 ⌬ 5/ 2 典 ⫽ A⌳共 2 ⌬兲 ⫽ ⫺a ␦
2 internuclear distance as compared with other observed low-
3 1
具 4 ⌸ 5/ 2 兩H SO兩 4 ⌸ 5/ 2 典 ⫽ A⌳共 4 ⌸兲 ⫽ ⫺a ␦ ⫺ a ␲ [3]
2 2
3 1
具 4 ⌽ 9/ 2 兩H SO兩 4 ⌽ 9/ 2 典 ⫽ A⌳共 4 ⌽兲 ⫽ ⫺a ␦ ⫹ a ␲ .
2 2
Copyright © 2000 by Academic Press
243
The a ␦ value of 380 cm ⫺1 which has been obtained from the
spin– orbit splitting of the X 5 ⌬ ground state of FeO (37) is
employed. We have deduced the a ␲ value by scaling the value
from the D 3 ⌸ state of VN (16). Taking the atomic spin– orbit
parameters of Fe and V (38), we obtain
a ␲ 共FeN兲 ⫽ a ␲ 共VN兲 ⫻ ␨ 共Fe兲/ ␨ 共V兲
[4]
⫽ 202.6 ⫻ 417/177 cm ⫺1 ⫽ 477 cm ⫺1 .
By substituting these values into Eq. [3], we have A⌳( 2 ⌬) ⫽
⫺760 cm ⫺1, A⌳( 4 ⌸) ⫽ ⫺412 cm ⫺1, and A⌳( 4 ⌽) ⫽ ⫺94
cm ⫺1. The regular spin– orbit separations of 4⌸ and 4⌽ states
are destroyed by the 4⌸– 6⌺, 4⌸– 2⌸, and 4⌽– 2⌽ interactions,
where the 2⌸ and 2⌽ states have the same configuration with
the 4⌸ and 4⌽ states. The 4⌸– 6⌺ interaction is considered to be
important because the calculated energy separation is very
small, where the off-diagonal elements can be estimated
as 具 4 ⌸ 5/ 2 兩H SO 兩 6 ⌺ 5/ 2 典 ⫽ a ␦ 具v 4 ⌸ 兩v 6 ⌺ 典, 具 4 ⌸ 3/ 2 兩H SO 兩 6 ⌺ 3/ 2 典 ⫽
(3/ 公15)a ␦ 具v 4 ⌸ 兩v 6 ⌺ 典, and so on using a pure precession
approximation of the ␦ and ␲ MO’s. Although less important,
the 4⌸– 2⌸ and 4⌽– 2⌽ interactions lower the energies of the
⌺ ⫽ ⫺ 21 and 12 components of the 4⌸ and 4⌽ states.
In this study, the ⍀ value of the ground state of FeN has
been determined to be 52, which should be the spin– orbit com-
ponent of 2⌬ or 4⌸. Theoretical studies suggest that 2⌬ is
slightly lower in energy than 4⌸, but we could not conclude
which is the ground state of FeN. One possible assignment is
that the ground state is 2⌬ 5/2, and the b and c states are the ⍀ ⫽
2 and 2 spin– orbit components of ⌸, respectively. The ⍀ ⫽ 2
5 3 4 3
⫺1
state at 21 514 cm (v⬘ ⫽ 1?) and the (V)⍀ ⫽ 2 (v⬘ ⫽ 1) state 5
at 21 377 cm ⫺1 have similar rotational constants and lifetimes.
These observations indicate that these states may be spin– orbit
components of the same electronic state. Considering the se-
lection rules of case(a)– case(a) transitions, the b and c states
are also spin– orbit components of the same electronic states.
The energy separation of 218 cm ⫺1 between the b and c states
seems too small for 4⌸ and too large for 4⌽ if we compare this
value with the expected spin– orbit separations. The assign-
ment that the b and c states are the ⍀ ⫽ 25 and 32 spin– orbit
components of 4⌸, respectively, might be plausible if the
perturbations discussed above are considered. It is noteworthy
that the interacting 6⌺ state was not identified in the DF
spectra.
The tentative assignment that the ground state is 2⌬ does not
conflict with the results of the Franck–Condon calculations
which suggest that the ground state seems to have the smaller
lying states. Now the question to be answered is what the a
state is. If the ground state is assigned as 2⌬, the a state is
considered to be a spin– orbit component of 4⌸ or 4⌽. The
Franck–Condon analyses estimated the r 0 value of the a state
to be greater than that of the X⍀ ⫽ 25 state by about 10%. This
244 AIUCHI AND SHIBUYA
value is too large for the 4⌸ or 4⌽ state because the 4␲ orbital
has only a slight antibonding character. This is partly due to
our inability to estimate the Franck–Condon factors accurately.
To rationalize these complicated electronic structures, more
experimental data are required.
The ⌬G 1/ 2 value of the X⍀ ⫽ 25 state is determined to be 870
cm ⫺1 in this study, but this value does not accord with the value
previously reported by Chertihin et al. (39). These authors
reported the matrix infrared spectra of the reaction products of
laser-ablated Fe atoms with molecular nitrogen, and the carrier
of the 938 cm ⫺1 band was assigned as FeN. The discrepancy
between these values is rather large for a matrix cage effect.
Possible explanations of this discrepancy might be that the
electronic state observed in this work is different from that in
the work of Chertihin et al., or that they observed other species.
In the LIF excitation spectra, the intensities of the (I)–
(V)⍀ ⫽ 25–X⍀ ⫽ 25 band systems are roughly in the order (V) ⬎
(IV) ⬇ (I) ⬇ (II) ⬎ (III). The fact that the DF spectra from the
(I) and (II) states show mainly the vibrational structure of the
X state indicates that the (I)–X and (II)–X transitions are
symmetrically allowed, and the (I) and (II) states have the same
symmetry with the X state. On the other hand, because no
vibrational structures of the X⍀ ⫽ 25 state were observed or
because these structures were weak in the DF spectra from the
(V) and (IV) states, it is natural to consider that the (V)–X and
(IV)–X transitions are symmetrically forbidden and borrow
intensity from other “allowed” transitions. Therefore, the (I)–X
and (II)–X transitions are much weaker than these allowed
transitions. One of the (I)–X and (II)–X transitions may be
ascribed to the transition in which one of the 9␴ electrons is
promoted to the 10␴ orbital. Fiedler and Iwata suggested that
the 9␴ is the hybrid orbital of the 4s and 3d ␴ , and the 10␴
orbital is the antibonding orbital between the sd ␴ hybrid orbital
and 2p ␴ (N) orbital, though it is strongly polarized toward the
metal center. The 9␴ 3 10␴ transition is expected to be weak
Copyright © 2000 by Academic Press
because it corresponds to the metal-centered s 3 d or d 3 d
transition. This transition is analogous to the D 5 ⌬–X 5 ⌬ tran-
sition of FeO (40) observed at 17 091 cm ⫺1. The r 0 value of the
excited state is greater by 5.0% than that of the ground state for
FeO, which is comparable to the values of 3.2% for the (I)–X
transition and 3.7% for the (II)–X transition, but the agreement
is not perfect. This indicates that the orbital characters change
on moving from FeO to FeN, or these considerations based on
a single-configuration picture are too simple for these mole-
cules. The r 0 values of the (III), (IV), and (V) states are greater
by about 14, 11, and 10% than that of the ground state,
respectively. These drastic geometry changes indicate that
these transitions are probably charge-transfer transitions in
which an electron mainly localized on the N atom is promoted
to an orbital mainly localized on the Fe atom.
7. CONCLUSION
In this study, the LIF excitation spectra of jet-cooled 56Fe 14N
and 56Fe 15N between 16 300 and 21 600 cm ⫺1 have been mea-
sured. All rotationally resolved transitions were found to orig-
inate from the ground ⍀ ⫽ 25 state, for which the most probable
candidate is the spin– orbit component of the 2 ⌬ i state arising
from the configuration of . . . 8␴ 23␲ 41␦ 39␴ 2. The X⍀ ⫽ 52
rotational constant was determined to be B 0 ⫽ 0.602793(17)
cm ⫺1, giving r 0 ⫽ 1.580 Å. The excited electronic states show
very complicated rovibronic structures due to a lot of pertur-
bations. The band systems from (I)⍀ ⫽ 25–X⍀ ⫽ 25 to (V)⍀ ⫽
2–X⍀ ⫽ 2 have been identified with the Franck–Condon pat-
5 5
terns of the DF spectra. The isotope shift was shown not to be
helpful to make the vibrational numbering for many bands. In
the DF spectra, three low-lying electronic states were observed
545, 1407, and 1625 cm ⫺1 above the X⍀ ⫽ 52 state. The
assignments of these states are unclear and must await further
experimental data.
 JET-COOLED OPTICAL SPECTROSCOPY OF FeN 245

APPENDIX 1

Assigned Rotational Lines of 56

Fe 14N in cm ⴚ1

Note. The symbol of * indicates blended or questionably assigned lines. The symbol of B indicates broadened lines. The numbers in parentheses are observed
minus calculated line positions in 10 ⫺3 cm ⫺1. These are listed only for lines included in the least-squares fit.

Copyright © 2000 by Academic Press

246 AIUCHI AND SHIBUYA

APPENDIX 1—Continued
JET-COOLED OPTICAL SPECTROSCOPY OF FeN 247

APPENDIX 1—Continued

Copyright © 2000 by Academic Press

APPENDIX 1—Continued
JET-COOLED OPTICAL SPECTROSCOPY OF FeN 249

APPENDIX 1—Continued

Copyright © 2000 by Academic Press

250 AIUCHI AND SHIBUYA

APPENDIX 1—Continued
JET-COOLED OPTICAL SPECTROSCOPY OF FeN 251

APPENDIX 1—Continued
252 AIUCHI AND SHIBUYA

APPENDIX 1—Continued

Copyright © 2000 by Academic Press

APPENDIX 1—Continued
254 AIUCHI AND SHIBUYA

APPENDIX 2

Assigned Rotational Lines of 56

Fe 15N in cm ⴚ1

Note. The symbol of * indicates blended or questionably assigned lines. The symbol of B indicates broadened lines. The numbers in parentheses are observed
minus calculated line positions in 10 ⫺3 cm ⫺1. These are listed only for lines included in the least-squares fit. For 56Fe 15N, high-resolution (0.04 cm ⫺1) spectra
were recorded for the bands at 19 608, 20 536, 21 344, 21 372, and 21 383 cm ⫺1. The other bands were recorded at a resolution of 0.2 cm ⫺1.
JET-COOLED OPTICAL SPECTROSCOPY OF FeN 255

APPENDIX 2—Continued
256 AIUCHI AND SHIBUYA

APPENDIX 2—Continued
JET-COOLED OPTICAL SPECTROSCOPY OF FeN 257

APPENDIX 2—Continued

Copyright © 2000 by Academic Press

258 AIUCHI AND SHIBUYA

APPENDIX 2—Continued

Copyright © 2000 by Academic Press

JET-COOLED OPTICAL SPECTROSCOPY OF FeN 259

APPENDIX 2—Continued

Copyright © 2000 by Academic Press

260 AIUCHI AND SHIBUYA

APPENDIX 2—Continued
 JET-COOLED OPTICAL SPECTROSCOPY OF FeN
ACKNOWLEDGMENTS
This work has been supported in part by a Grant-in-Aid from the Ministry
of Education, Science, Sports and Culture (Contracts 09440196 and
11131209).
REFERENCES
1. H. Spinrad and R. F. Wing, Annu. Rev. Astron. Astrophys. 7, 249 –302
(1969).
2. P. W. Merrill, A. J. Deutsch, and P. C. Keenan, Astrophys. J. 136, 21–34
(1962).
3. A. J. Sauval, Astron. Astrophys. 62, 295–298 (1978).
4. R. Yerle, Astron. Astrophys. 73, 346 –351 (1979).
5. M. Grunze, in “The Chemical Physics of Solid Surfaces and Heteroge-
neous Catalysis” (D. A. King and D. P. Woodruff, Eds.), pp. 143–194,
Elsevier, New York, 1982.
6. R. S. Ram and P. F. Bernath, J. Chem. Phys. 96, 6344 – 6347 (1992).
7. T. M. Dunn, L. K. Hanson, and K. A. Rubinson, Can. J. Phys. 48,
1657–1663 (1970).
8. J. K. Bates, N. L. Ranieri, and T. M. Dunn, Can. J. Phys. 54, 915–916
(1976).
9. A. E. Douglas and P. M. Veillette, J. Chem. Phys. 72, 5378 –5380 (1980).
10. C. Athénour and J.-L. Féménias, Can. J. Phys. 60, 109 –116 (1982).
11. K. Brabaharan, J. A. Coxon, and A. B. Yamashita, Can. J. Phys. 63,
997–1004 (1985).
12. B. Simard, H. Niki, and P. A. Hackett, J. Chem. Phys. 92, 7012–7020
(1990).
13. D. A. Fletcher, C. T. Scurlock, K. Y. Jung, and T. C. Steimle, J. Chem.
Phys. 99, 4288 – 4299 (1993).
14. B. Simard, C. Masoni, and P. A. Hackett, J. Mol. Spectrosc. 136, 44 –55
(1989).
15. S. L. Peter and T. M. Dunn, J. Chem. Phys. 90, 5333–5336 (1989).
16. W. J. Balfour, A. J. Merer, H. Niki, B. Simard, and P. A. Hackett, J. Chem.
Phys. 99, 3288 –3303 (1993).
17. T. C. Steimle, J. S. Robinson, and D. Goodridge, J. Chem. Phys. 110,
881– 889 (1999).
Copyright © 2000 by Academic Press
261
18. W. J. Balfour, C. X. W. Qian, and C. Zhou, J. Chem. Phys. 106, 4383–
4388 (1997).
19. C. Zhou, W. J. Balfour, and C. X. W. Qian, J. Chem. Phys. 107, 4473–
4482 (1997).
20. G. H. Jeung and J. Koutecký, J. Chem. Phys. 88, 3747–3760 (1988).
21. K. L. Kunze and J. F. Harrison, J. Am. Chem. Soc. 112, 3812–3825 (1990).
22. K. D. Carlson and C. R. Claydon, J. Chem. Phys. 46, 4963– 4969 (1967).
23. C. W. Bauschlicher, Jr., Chem. Phys. Lett. 100, 515–518 (1983).
24. S. M. Mattar, J. Phys. Chem. 97, 3171–3175 (1993).
25. S. M. Mattar, W. D. Hamilton, and C. T. Kingston, Chem. Phys. Lett. 271,
125–132 (1997).
26. J. F. Harrison, J. Phys. Chem. 100, 3513–3519 (1996).
27. S. M. Mattar and B. J. Doleman, Chem. Phys. Lett. 216, 369 –374 (1993).
28. M. R. A. Blomberg and P. E. M. Siegbahn, Theor. Chim. Acta 81, 365–374
(1992).
29. P. E. M. Siegbahn and M. R. A. Blomberg, Chem. Phys. 87, 189 –201 (1984).
30. A. Fiedler and S. Iwata, Chem. Phys. Lett. 271, 143–151 (1997).
31. A. Daoudi, A. T. Benjelloun, J. P. Flament, and G. Berthier, J. Mol.
Spectrosc. 194, 8 –16 (1999).
32. R. S. Ram, J. Liévin, and P. F. Bernath, J. Chem. Phys. 109, 6329 – 6337
(1998).
33. R. S. Ram, J. Liévin, and P. F. Bernath, J. Chem. Phys. 111, 3449 –3456
(1999).
34. J. Cariou and P. Luc, “Atlas du spectre d’absoption de la molecule de
Tellure.” CNRS, Paris, 1980.
35. S. Gerstenkorn and P. Luc, “Atlas du Spectre d’Absoption de la Molecule
d’Iode,” CNRS, Paris, 1978.
36. G. Herzberg, “Molecular Spectra and Molecular Structure. I. Spectra of
Diatomic Molecules,” p. 456, Van Nostrand–Reinhold, New York, 1950.
37. M. D. Allen, L. M. Ziurys, and J. M. Brown, Chem. Phys. Lett. 257,
130 –136 (1996).
38. H. Lefebvre-Brion and R. W. Field, “Perturbations in the Spectra of
Diatomic Molecules.” Academic Press, Orlando, FL, 1986.
39. G. V. Chertihin, L. Andrews, and M. Neurock, J. Phys. Chem. 100,
14609 –14617 (1996).
40. M. Barnes, M. M. Fraser, P. G. Hajigeorgiou, A. J. Merer, and S. D.
Rosner, J. Mol. Spectrosc. 170, 449 – 465 (1995).