1

Journal of Molecular Spectroscopy 224, 137 (2004)

The Microwave Spectrum of the FeS Radical

Shuro Takano,* Satoshi Yamamoto,† and Shuji Saito‡ *Nobeyama Radio Observatory1, and Department of Astronomical Science, The Graduate University for Advanced Studies (Sokendai), Minamimaki, Minamisaku, Nagano, 384-1305 Japan; †Department of Physics, The University of Tokyo, Hongo, Bunkyo, Tokyo, 113-8654 Japan; ‡Research Center for Development of Far-Infrared Region, Fukui University, Bunkyo, Fukui, 910-8507 Japan

ABSTRACT

The rotational spectrum of the iron monosulfide radical, FeS, was measured in the frequency

region of 220 to 390 GHz with a source-modulated millimeter/submillimeter-wave spectrometer.

The radical was efficiently produced in a free space absorption cell by a dc discharge in a mixture of

Ar and H2S with a stainless-steel hollow cathode. Several series of paramagnetic lines were

detected with intervals of about 12 GHz. The four series having relatively strong intensity were

assigned to FeS in the vibrationally ground state of the X5∆i electronic state, two series to that in the

vibrationally excited state, and five series presumably to FeS in the electronically excited state, A5Σ+.

The effective molecular constants were determined for FeS in the X5∆i electronic state. The Ω=2

components of the vibrationally ground state showed an apparent shift from the typical pattern of

the 5∆i state. In addition, the fine structure of the A5Σ+ state was found to be far from a regular

pattern expected for a 5Σ state. A trial analysis including electronic interaction between the 5∆i and 5Σ states was carried out, but it was not possible to explain the spectral lines of both electronic states

simultaneously. Reasons for the heavily perturbed spectral patterns are discussed.

Keywords: microwave spectrum, free radical, iron monosulfide, perturbation

1 Nobeyama Radio Observatory is a branch of the National Astronomical Observatory, an inter-university research institute operated by the Ministry of Education, Culture, Sports, Science and Technology.

2

INTRODUCTION

Several transition metal oxides have been studied by microwave spectroscopy: e.g., TiO (1),

MnO (2), FeO (3, 4), CoO (5), NiO (6), CuO (7), AgO (8), and ZrO (9). Detailed molecular constants

in multiplet ground electronic states give much new information on the nature of the metal oxide

bond and various interactions within the state and between the electronic states. Transition metal

sulfides have been less studied by high-resolution spectroscopy than corresponding oxides. When

oxygen is replaced by sulfur, substantial change is expected in bonding between transition metal

and sulfur and in the complicated electronic structure, because sulfur has smaller electronegativity

than oxygen, and the 3p of sulfur has larger distribution than the 2p of oxygen, changing s-d-p

hybridization in the bonding.

Among transition metal oxides listed above, FeO has been well studied for ages by optical

spectroscopy, which established the ground electronic state of FeO to be X5∆i (10, 11). Its molecular

properties were determined in details by various methods such as microwave spectroscopy (3, 4),

molecular beam laser spectroscopy (12), MODR spectroscopy (13), LIF spectroscopy (14, 15), and

photoelectron spectroscopy (16). Optical spectroscopy and photoelectron spectroscopy gave evidences

on the existence of low-lying electronic states, A5Σ+ (16) and a7 Σ + (14). Contrary to FeO, there are

few spectroscopic studies of iron sulfide. DeVore and Franzen (17) studied vibrational spectra of

FeS in Ar and OCS matrices, and Zhang et al. (18) measured photoelectron spectra. Few

quantum chemical calculations are available for its bond length (re), harmonic frequency (ωe),

dissociation energy (D0), and dipole moment (µ) (19, 20, 21). Hübner et al. (20), who report

high-level calculations for the electronic energy levels of FeS as well as an interpretation of the

photoelectron spectrum (18), strongly suggest that its state order near the ground state is 5Σ+ (0.0

eV), 5∆ (0.13 eV), and 7Σ+ (0.26 eV).

Iron bearing molecules are interesting for their possible existence in stellar and interstellar

sources, because iron is the most stable element produced in the nuclear syntheses in stars and it

has the highest cosmic abundance among heavier metals, comparable to that of silicon. Several

silicon-bearing molecules were detected in circumstellar envelopes and star forming regions. The

rotational spectral lines of FeO were searched for toward stars and even toward interstellar clouds

since their line frequencies were measured in the laboratory (3). Only upper limits to the

abundance were reported (22, 23). In cold interstellar space the iron bearing molecules are

significantly depleted in dust grains, but they may exist in the gas phase in hot regions such as

circumstellar envelopes of late-type stars and star forming regions. Very recently Walmsley et al.

(24) reported a tentative detection of FeO. Its J=5-4 line was observed in absorption at 153 GHz

toward the hot and relatively low-density gas in the region of Sgr B2. This absorption line was

further confirmed by interferometric observations toward the same object (25). Tsuji (26) predicted

that FeS is more abundant than FeO in cool stellar atmospheres. However, no searches have been

3

carried out for FeS, because no spectral line frequencies are available.

In the present study, we observed the rotational spectrum of FeS in the ground electronic state

of X5∆i for the first time in the laboratory, and also detected spectral lines which were presumably

assigned to a low-lying electronically excited state of A5Σ+.

EXPERIMENTAL

The 100 kHz source-modulated microwave spectrometer of Nagoya University was used to

measure spectral lines of FeS (27). For production of metal-bearing molecules in the gas phase, the

dc sputtering method was employed: a mixture of Ar (or He) and H2S (or OCS, CS2) was discharged

in a 2 m long free space absorption cell equipped with a 1.4 m long stainless steel hollow cathode.

Iron atoms were supplied from the surface of the stainless steel hollow cathode to the gas phase.

Since no theoretical prediction was available for FeS at the time of the experiments, the rotational

constant of FeS was estimated to be between 5.1 and 6.3 GHz from bond lengths of related

molecules. The ground electronic state was assumed to be 5∆i, as suggested from FeO, and

paramagnetic lines were searched for in a wide frequency range of more than 20 GHz.

Consequently, several series of paramagnetic lines were found as schematically shown in Figure 1,

where the frequency axis is given with an interval of 12 GHz, close to the interval of two adjacent

lines of each series. The interval is close to two times of the expected rotational constants for FeS.

The lines indicated by closed circles show relatively large Zeeman effect: The line width (FWHM)

increases from ~0.7 MHz to ~2.0 MHz as the magnetic field applied to the cell is increased from zero

to several Gauss. This is in sharp contrast to other series of lines, where the line width (FWHM)

increases to ~2.0 MHz only at the magnetic field of more than 50 Gauss. The different behavior

against the magnetic field is, therefore, rather striking. Two series showing a doublet pattern were

indicated in Figure 1. These doublings were considered to be Λ–type doubling.

These paramagnetic lines changed in intensity with production conditions. (1) The intensity

decreased as a trace amount of oxygen was introduced into the cell. (2) The lines were observable

even by discharging Ar (He) with only a trace amount of H2S (or OCS, CS2). (3) The intensity

decreased severely when the discharge current changed from 500 mA to 50 – 100 mA. The

optimum condition for these lines was obtained at a discharge current of 500 mA in a mixture of Ar

(1.3 Pa or He, 2.7 Pa) and H2S (or OCS, CS2, 0.1 Pa). The intensity increased as the current

increased, but, for maintaining stable discharge, a discharge current of 500 mA was used

throughout measurements. Change in the cell temperature did not affect much the production,

and the temperature was kept between 190 and 270 K.

Several lines of series with relatively strong intensity were investigated to see whether they

were due to an iron bearing free radical. On replacing the stainless-steel hollow cathode by an

aluminum electrode, they disappeared. Subsequently, ferrocene, (C5H5)2Fe, was introduced into

4

the cell with the aluminum hollow cathode, and a discharge in an addition of OCS reproduced

weakly the paramagnetic lines. Therefore, we concluded that these lines are due to an iron and

sulfur bearing free radical, most likely FeS. An example of the observed spectral lines is shown in

Figure 2, which was observed at 387.7 GHz and later assigned to the J=32-31 transition of FeS in

the X5∆4 substate.

ANALYSIS

First, each line frequency of the series was divided by a tentatively assumed rotational quantum

number for the upper level, and the derived value (two times the effective rotational constant) was

plotted against the rotational quantum number as shown in Figure 3 (a) and (b). The derived

effective rotational constant shows a harmonic relation with the quantum number (Figure 3 (a))

except for the five series indicated by closed circles in Figure 1 (Figure 3 (b)). As a result, the series

given in Figure 1 are classified into two groups: six series with harmonic relations and five series

without harmonic relations against assumed quantum numbers.

The 5∆i state

Two weak series among six ones with harmonic relations show doublet structures suggesting

Λ–type doubling, and two series at the lower frequency side are considered to belong to the v=1 state

from their weaker intensity. For the assignment of the Ω substate, doubling separations were

examined for the two series. When the doubling separation was plotted against the rotational

quantum number, the separation for the lower frequency series was found to be roughly

proportional to J4 and that for the upper frequency series approximately to J1.5. According to

Brown, Cheung, and Merer (28), the Λ–type doubling separation is nearly proportional to

[J(J+1)]Ω /J ∼ J2Ω- 1 for the ∆ state. Therefore, the Ω values were calculated to be 2.5 and 1.3 for

the lower and upper frequency series, respectively. These features suggest that the observed

spectral lines are due to a molecule in a degenerate electronic state with two or more electronic spin

quantum number and with a negative spin-orbit coupling constant, that is, inverted. Thus, these

series with harmonic relations were assigned to the spectral lines of FeS in the v=0 and 1 states of

the 5∆i electronic state, as the ground electronic state of FeO suggested. The Ω value assigned to

the fine structure component is indicated in Figure 1. The spectral lines of the Ω=0 substate were

not clearly identified in this study, because they should be very weak and might have irregular

pattern due to perturbation.

The spectral lines assigned to the X5∆i electronic state were analyzed by the least-squares

method using the matrix elements for a Hamiltonian appropriate to the 5∆ electronic state (14, 28).

At first the spectral lines of the Ω =4, 3, 2, and 1 substates were analyzed simultaneously, but those

of the Ω =2 substate showed systematic residuals, which were significantly larger than the

5

frequency measurement errors. Therefore, the spectral lines of the Ω =2 substate were not

included in the final fit. The correlations among the parameters were rather high only between

ASO and AD, but both of them could be determined. The spectral lines of the v=1 state were

analyzed similarly where B, D, and ASO were employed as floating parameters, because only the

spectral lines of the Ω =4 and 3 substates were observed for the v=1 state. The correlations among

the parameters were not so high to prevent to determine the parameters. The observed and

calculated line frequencies for the v=0 state are listed in Table 1, and those for v=1 in Table 2. The

determined molecular constants are listed in Table 3.

The 5Σ+ state

As noted above, the effective rotational constants for the five series indicated by closed circles

in Figure 1 show no harmonic relation against the assumed quantum number. Since every set of

the five lines have similar intensity (but significantly weaker than those of the 5∆4 and 5∆3 states in

the vibrationally ground state, see Figure 1) and no other series lines with comparable intensities

are observed around the five lines, we conclude that these five series lines are due to FeS in the A5Σ+

state, as suggested from the excited electronic state of FeO (16). The observed spectral lines for

these series were analyzed using a Hamiltonian for the 5Σ state (29). However, it was not

straightforward to assign each line to the component of the 5Σ state, and after several trial

calculations the observed pattern was roughly reproduced with a set of molecular constants:

Beff=6227.6 MHz, λ= –103000 MHz, γ= –733 MHz, and θ=17500 MHz. The standard deviation of

the least-squares fit was about 100 MHz, which was far from the measurement errors of observed

lines. Since the molecular constants derived are so poor to explain the observed spectral lines that

only a tentative assignment is given for each spectral line as listed in Table 4.

DISCUSSION

Production of FeS with sputtering method

In the present study the relatively poorly studied transition metal sulfide, FeS, was

characterized for the first time by using a high-sensitivity millimeter/submillimeter-wave

spectrometer combined with the dc sputtering production method (30). Nonvolatile heavy metal

bearing molecules have been usually produced in the gas phase with high-temperature ovens such

as heat-pipe oven or King furnace. The high-temperature oven evaporates heavy metal directly in

the gas phase and produced molecules remain at high temperature in general. On the other hand,

the sputtering method produces gaseous metal containing molecules at relatively low temperature.

This feature of the sputtering method is adequate for microwave spectroscopy without a very high

temperature cell, and has been applied to spectroscopy of several transition metal oxides (2, 5-8, 31,

32).

6

Electronic states of FeS

According to the photoelectron spectroscopy of FeS (18), the order and energ ies of the

electronic states are as follows: 0 eV (5∆), 0.49 eV (5Σ+), 0.66 eV (3Σ+), and 1.03 eV ( 5∆), where they

thought that the ground electronic state of FeS– was 4∆. On the other hand, the detailed ab initio

calculations (CASSCF/ACPF) (20) predicted that the ground electronic state of FeS– was 6∆. They

reassigned the bands obtained with the photoelectron spectroscopy based on their calculations.

The order and energies of the electronic states of FeS obtained were 0 – 0.1 eV (5Σ+, 5∆, 7Σ+), 0.49 eV

(5Π, 5Φ), 0.66 eV (7Π, 7Φ), and 1.03 eV (5Σ–, 7∆). Therefore, there are many electronic states in FeS

in low energy. Our experiment of microwave spectroscopy indicates that the 5∆ and 5Σ+ states

actually exist in low energy. As mentioned before, their properties are rather different from each

other in the aspects of Zeeman effect and harmonic relation with respect to the rotational quantum

number. These different properties indicate that the mixing is not severe between the 5∆ and 5Σ+

states. Based on the observed line intensities, the 5∆ state is most probably the electronic ground

state. We could not find other states with comparable intensity. If several electronic states of FeS

actually exist in quite low energy as predicted by the above-mentioned ab initio calculations, the 5∆

and 5Σ+ states, which we observed, near (or at) the ground state are probably severely perturbed

each other and by other nearby states. In this case the properties of the 5∆ and 5Σ+ states should be

similar due to severe mixing of the states by perturbations, and the Ω value assignments may have

little sense. However, we conclude that the perturbations are not so severe as to prevent the Ω

value assignments based on the results of our measurements.

Molecular constants

Since the molecular constants of FeS (5∆) were determined spectroscopically for the first time, it

may be worthy to compare them with those predicted by an ab initio calculation. The observed

rotational constant B0 of 6106.16(10) MHz is well compared with the predicted value of 6058 MHz

for Be, which is calculated from re of 2.025 Å (CASSCF) (19). Almost the same re of 2.03 – 2.04 Å

(CASSCF, DFT-B3LYP) is obtained by other ab initio calculation (20).

The same calculation above (19) predicts the harmonic frequency of FeS ωe to be 521 cm-1, which

is nearly equal to the fundamental frequency ω0 of 506 cm-1, derived from the present centrifugal

distortion constant D0 (28). Other ab initio calculation (20) predicts ωe to be 519 cm-1

(CASSCF/ACPF) and ~530 cm-1 (DFT-B3LYP). In addition, DeVore and Franzen (17) reported ω0

of FeS in OCS and Ar matrices to be 539 and 542 cm-1, respectively.

The spin-orbit coupling constant ASO was determined to be about –45 cm-1. Although this is an

effective value, it is nearly half of that of FeO (14). Generally sulfides have larger spin-orbit

coupling constants than corresponding oxides. For example, CuO (X2Πi) has the ASO of –276.16

7

cm-1 (34) and CuS (X2Πi) that of –433.20 cm-1 (35). The small effective spin-orbit coupling constant

in the present study implies that not only the 5∆2 sublevel but the whole 5∆ state is perturbed. The

spin-orbit coupling constant of the v=1 state is significantly different from that of the v=0 state.

This is also the effect of perturbation, and an effective spin-orbit coupling constant is derived.

The Ω =2 substate lines in the X5∆i state

The Ω =2 component of the 5∆i state show large residuals in the least-squares fit. The

Λ–type doubling separation is abnormally large, and does not obey the typical pattern of the 5∆i state,

as noted above. The low-lying Σ electronic states are the most likely candidates as the perturbing

state. The ∆ and Σ electronic states interact through the spin-spin interaction. According to the

selection rule of ∆S=0, ±1, ±2, ∆Ω=0, ∆Λ=±2, and ∆Σ= m2 for the spin-spin interaction (36), the most

possible perturbing state was considered to be A5Σ+ or a7Σ+. As a trial analysis, a least-squares

program including both the 5∆ and 5 Σ states and the homogeneous spin-spin coupling term between

them was developed and used. However, the inclusion of the electronic interaction between the two

states did not improve the residuals of the 5 Σ state. We wonder that the A5 Σ+ state also suffers

from strong perturbations from a nearby electronic state, for an example, a7 Σ state. So as to

analyze perturbations in the A5 Σ + state as well as in the Ω =2 substate of the 5∆ state, further

detailed information for relative energy differences of the nearby electronic states and their

molecular constants is highly desirable.

Still assignments of part of series lines and analysis of perturbation have ambiguity, because

we cannot have a definite image of the perturbers. For further analysis detections and

measurements in microwave spectroscopy of the additional series lines such as in the Ω=0 state, in

the v=2 state, and of the 54FeS isotopic species (54FeS/56FeS ∼ 1/16) will also be very helpful. In

addition, direct measurements of electronic transitions mainly by optical spectroscopy will be

important for unraveling the electronic states.

Astronomical search

In the circumstellar envelope of the late-type carbon star, IRC+10216, the abundance of SiS is

higher than that of SiO by a factor of 10 times or more (37, 38). This is because the silicon atom

first combines with the oxygen atom, forming SiO, which is more stable than SiS, and the residual

silicon produces a larger amount of SiS in the less oxygen-abundant carbon star (30). The thermal

equilibrium calculation for carbon star predicts a possibility that the abundance of metal sulfides is

larger than that of the corresponding oxides in the relatively cool stellar envelope of the carbon star

(39). Hence, FeS may be detected in the carbon star envelope.

The spectral lines of the J=7-6 and 8-7 transitions in the 5∆4 substate, which is the lowest

substate, were searched for toward one carbon star envelope, IRC+10216 and three giant molecular

8

clouds including hot star-forming region, Orion-KL, Sgr B2, and W51. The 45 m radiotelescope of

the Nobeyama Radio Observatory was used for the observations. However, lines corresponding to

FeS were not detected at the noise level of 11 – 29 mK in antenna temperature.

Acknowledgment

We thank the staff members of the 45 m radiotelescope of the Nobeyama Radio Observatory for

their support during the observations. We also thank an anonymous referee for helpful comments.

REFERENCES

1. K. Namiki, S. Saito, J. S. Robinson, T. C. Steimle, J. Mol. Spectrosc. 191 (1998) 176-182.

2. K. Namiki, S. Saito, J. Chem. Phys. 107 (1997) 8848-8853.

3. Y. Endo, S. Saito, E. Hirota, Astrophys. J. 278 (1984) L131-L132.

4. M. D. Allen, L. M. Ziurys, J. M. Brown, Chem. Phys. Letters 257 (1996) 130-136.

5. K. Namiki, S. Saito, J. Chem. Phys. 114 (2001) 9390-9394.

6. K. Namiki, S. Saito, Chem. Phys. Letters, 252 (1996) 343-347.

7. T. Steimle, K. Namiki, S. Saito, J. Chem. Phys. 107 (1997) 6109-6113.

8. T. Steimle, M. Tanimoto, K. Namiki, S. Saito, J. Chem. Phys. 108 (1998) 7616-7622.

9. S. A. Beaton, M. C. L. Gerry, J. Chem. Phys. 110 (1999) 10715-10724.

10. A. S-C. Cheung, R. M. Gordon, A. J. Merer, J. Mol. Spectrosc. 87 (1981) 289-296.

11. A. S-C. Cheung, N. Lee, A. M. Lyyra, A. J. Merer, A. W. Taylor, J. Mol. Spectrosc. 95 (1982) 213-225.

12. T. Kröckertskothen, H. Knöckel, E. Tiemann, Chem. Phys. 103 (1986) 335-343.

13. T. Kröckertskothen, H. Knöckel, E. Tiemann, Mol. Phys. 62 (1987) 1031-1040.

14. A. W. Taylor, A. S-C. Cheung, A. J. Merer, J. Mol. Spectrosc. 113 (1985) 487-494.

15. T. C. Steimle, D. F. Nachman, J. E. Shirley, J. Chem. Phys. 90 (1989) 5360-5363.

16. G. Drechsler, U. Boesl, C. Bäßmann, E. W. Schlag, J. Chem. Phys. 107 (1997) 2284-2291.

17. T. C. DeVore, H. F. Franzen, High Temp. Sci. 7 (1975) 220-235.

18. N. Zhang, T. Hayase, H. Kawamata, K. Nakao, A. Nakajima, K. Kaya, J. Chem. Phys. 104 (1996) 3413-3419.

19. C. W. Bauschlicher, Jr., P. Maitre, Theor. Chim. Acta, 90 (1995) 189-203.

20. O. Hübner, V. Termath, A. Berning, J. Sauer, Chem. Phys. Lett. 294 (1998) 37-44.

21. O. Hübner, J. Sauer, Phys. Chem. Chem. Phys. 4 (2002) 5234-5243.

22. A.J. Merer, C.M. Walmsley, E. Churchwell, Astrophys. J. 256 (1982) 151-155.

23. J. Cernicharo, M. Guélin, Astron. Astrophys. 183 (1987) L10-L12.

24. C. M. Walmsley, R. Bachiller, G. Pineau des Forêts, P. Schilke, Astrophys. J. 566 (2002) L109-L112.

25. R. S. Furuya, C. M.. Walmsley, K. Nakanishi, P. Schilke, R. Bachiller, Astron. Astrophys. 409 (2003) L21-L24.

26. T. Tsuji, Astron. Astrophys. 23 (1973) 411-431.

27. S. Yamamoto, S. Saito, J. Chem. Phys. 89 (1988) 1936-1944.

9

28. J. M. Brown, A. S-C. Cheung, A. J. Merer, J. Mol. Spectrosc. 124 (1987) 464-475.

29. A. S-C. Cheung, W. Zyrnicki, A. J. Merer, J. Mol. Spectrosc. 104 (1984) 315-336.

30. S. Takano, Ph.D. Thesis (Nagoya University, 1991).

31. M. Goto, S. Takano, S. Yamamoto, H. Ito, S. Saito, Chem. Phys. Lett. 227 (1994) 287-292.

32. M. Tanimoto, S. Saito, T. Okabayashi, Chem. Phys. Lett. 242 (1995) 153-156.

33. C. H. Townes, A. L. Schawlow, “Microwave Spectroscopy”, chap. 1, Dover, New York, 1975.

34. L. C. O’Brien, R. L. Kubicek, S. J. Wall, D. E. Koch, R. J. Friend, C. R. Brazier, J. Mol. Spectrosc. 180 (1996)

365-368.

35. L. C. O’Brien, M. Dulick, S. P. Davis, J. Mol. Spectrosc. 195 (1999) 328-331.

36. H. Lefebvre-Brion, R. W. Field, “Perturbations in the Spectra of Diatomic Molecules”, chap. 2, Academic Press,

Orlando, 1986.

37. R. J. Boyle, J. J. Keady, D. E. Jennings, K. L. Hirsch, G. R. Wiedemann, Astrophys. J. 420 (1994) 863-868.

38. J. J. Keady, S. T. Ridgway, Astrophys. J. 406 (1993) 199-214.

39. T. Tsuji, private communication (1990).

10

TABLE 1

Observed and calculated transition frequencies of

the FeS radical in the v=0 state (X5∆i) (MHz)

J’ – J Ω νobs.a νobs. –νcalc.

19 – 18 4 230383.569(17) -0.004

20 – 19 4 242497.470(26) 0.000

21 – 20 4 254609.589(7) -0.001

22 – 21 4 266719.857(22) 0.015

23 – 22 4 278828.119(7) -0.020

24 – 23 4 290934.432(15) 0.043

25 – 24 4 303038.498(14) -0.008

26 – 25 4 315140.389(7) -0.009

27 – 26 4 327239.984(13) 0.006

28 – 27 4 339337.110(15) -0.046

29 – 28 4 351431.877(28) 0.035

30 – 29 4 363523.901(29) -0.047

31 – 30 4 375613.419(14) 0.035

32 – 31 4 387700.062(19) 0.002

19 – 18 3 230980.325(21) -0.046

20 – 19 3 243125.506(23) 0.013

21 – 20 3 255268.847(20) 0.040

22 – 21 3 267410.235(10) 0.010

23 – 22 3 279549.608(16) -0.046

24 – 23 3 291687.058(26) 0.052

25 – 24 3 303822.226(31) 0.037

26 – 25 3 315955.100(11) -0.013

27 – 26 3 328085.671(6) -0.016

28 – 27 3 340213.814(7) -0.006

29 – 28 3 352339.416(41) -0.007

30 – 29 3 364462.341(41) -0.063

31 – 30 3 376582.644(7) -0.028

32 – 31 3 388700.210(12) 0.072

J’ – J Ω νobs.a νobs. –νcalc.

18 – 17 2 219664.503(61) 130.011b

2 219666.863(47) 135.199

19 – 18 2 231859.268(55) 139.100

2 231862.157(29) 145.306

20 – 19 2 244052.544(21) 148.435

2 244056.082(17) 155.829

21 – 20 2 256244.535(21) 158.311

2 256248.678(42) 166.903

22 – 21 2 268434.889(19) 168.467

2 268439.870(37) 178.547

23 – 22 2 280623.677(35) 179.068

2 280629.571(29) 190.767

24 – 23 2 292810.836(19) 190.142

2 292817.794(41) 203.671

25 – 24 2 304996.254(17) 201.668

2 305004.388(23) 217.200

26 – 25 2 317179.928(22) 213.736

27 – 26 2 329361.779(22) 226.359

28 – 27 2 341541.789(50) 239.612

2 341554.688(35) 262.774

29 – 28 2 353719.841(26) 253.470

2 353734.839(23) 279.818

30 – 29 2 365895.983(19) 268.072

2 365913.371(25) 297.967

31 – 30 2 378070.080(22) 283.377

2 378090.227(25) 317.257

32 – 31 2 390242.182(10) 299.528

2 390265.469(41) 337.844

11

(table 1 continued)

J’ – J Ω νobs.a νobs. –νcalc.

28 – 27 1 342625.6 c 0.082

1 342719.2 c -0.148

29 – 28 1 354834.7 c -0.047

1 354933.6 c 0.239

30 – 29 1 367041.0 c -0.097

1 367144.6 c -0.035

31 – 30 1 379244.5 c 0.030

1 379353.0 c -0.074

32 – 31 1 391444.8 c 0.033

1 391558.6 c 0.017

a Values in parentheses indicate one standard

deviation of the frequency measurements

in kHz. b Weight is 0.0 for all Ω=2 transitions.

c Roughly measured frequency.

12

TABLE 2

Observed and calculated transition frequencies of

the FeS radical in the v=1 state (X5∆i) (MHz)

J’ – J Ω νobs. a νobs. –νcalc.

27 – 26 4 325712.7 b -0.101 c

28 – 27 4 337753.3 b -0.089 c

29 – 28 4 349791.530(30) 0.046

30 – 29 4 361826.971(5) -0.025

31 – 30 4 373859.856(14) 0.020

32 – 31 4 385889.7 b -0.216 c

27 – 26 3 326562.5 b -0.215 c

28 – 27 3 338634.390(26) 0.013

29 – 28 3 350703.523(14) 0.021

30 – 29 3 362769.976(9) -0.024

31 – 30 3 374833.785(34) 0.006

32 – 31 3 386894.8 b 0.050 c

a Values in parentheses indicate one standard deviation

of the frequency measurements in kHz.

b Roughly measured frequency. c Weight is 0.1.

13

TABLE 3

The molecular constants of the FeS radical in the X5∆i electronic state (MHz) a

Constants v=0 (Ω=4,3,1) v=1 (Ω=4,3)

ASO – 1340000(31000) – 2245100(230)

B 6106.16(10) 6069.416(10)

D 0.0039529(35) 0.0038465(58)

λ – 47000(5100) ---

λD – 0.232(79) ---

AD 2.38(33) ---

ñ∆ 10.73(29) ---

õ∆ 1.053(55) ---

a Values in parentheses are three standard deviation and apply

to the last digits of the constants.

14

TABLE 4

Observed frequencies and tentative assignments of the FeS

spectral lines in the A5Σ+ state (MHz) a

N’ – N F1 F2 F3 F4 F5

17 – 16 223176.888(3)

18 – 17 234912.782(12)

19 – 18 246649.625(8)

20 – 19 258396.986(23)

21 – 20 270162.747(22)

22 – 21 281953.151(24)

23 – 22 293772.992(6) 277527 b

24 – 23 305625.384(4) 301097.8 b 295881.5 b 298281.1 b

25 – 24 317512.039(23)

26 – 25 329433.684(44) 325628.1 b 321089.7 b 323004.3 b

27 – 26 341389.943(79)c 337905.6 b 333671.2 b 335369.7 b

28 – 27 353379.604(24) 350189.880(22) 342916.4 b 346238.598(16) 347736.898(34)

29 – 28 365401.285(16) 362480.421(9) 355859.4 b 358792.332(8) 360105.104(35)

30 – 29 377452.879(16) 374776.380(18) 368757.9 b 371333.321(13) 372473.982(11)

31 – 30 389532.251(14) 387076.9 b 381613.707(45) 383861.6 b 384843.09 b

32 – 31 394428.2 b

a Values in parentheses indicate one standard deviation of the frequency measurements in kHz. b Roughly measured frequency.

c Disturbed.

15

Figure 1. The spectral lines of FeS are shown schematically. The abscissa indicates observed

frequency regions with the interval of 12 GHz. The ordinate corresponds to relative intensity,

which has an error of about 10-20%. The lines indicated by closed circle show relatively large

Zeeman effect and were assigned to the lines due to a low-lying Σ electronic state. The lines

indicated by asterisk or open square are not well established to make series, and are not yet

assigned.

16

Figure 2. The J=32-31 rotational spectral line of FeS in the 5∆4 substate at 387.7 GHz. The

radical was produced by 500 mA dc discharge with a stainless-steel hollow cathode

electrode in a mixture of H2S (0.1 Pa) and Ar (1.3 Pa). The integration time is about

12 seconds with time constant of 1 ms for the lock-in amplifier.

17

Figure 3 (a)

Figure 3. Each line frequency of the series was divided by rotational quantum number for the

upper level, and the derived value (two times the effective rotational constant) was

plotted against the rotational quantum number. (a) Plot related to the 5∆ state.

The derived value shows a harmonic relation with the quantum number. The series

indicated by asterisk or open square are the same as those indicated by the same

marks in Figure 1. (b) Plot of the five series in the 5Σ state indicated by closed

circles in Figure 1. The derived value does not show a harmonic relation with the

quantum number.

18

Figure 3 (b)