Electronic spectroscopy of jet-cooled YbNH 3

We report the first spectroscopic study of a complex consisting of a rare earth atom in combination with ammonia. Using two-color resonance-enhanced multiphoton ionization (REMPI) spectroscopy, the lowest energy electronic transition of YbNH3 has been found in the near-infrared. The spectrum arises from a spin-forbidden transition between the A1 ground electronic state and the lowest 3E excited electronic state. The transition is metal centered and approximately correlates with the Yb 6s6p 3P ← 6s2 1S transition. The observation of clear spin-orbit structure in the spectrum confirms the C3v symmetry of YbNH3. Vibrational structure is also observed in the REMPI spectrum, which is dominated by excitation of the Yb–N stretching vibration. © 2012 American Institute of Physics. [http://dx.doi.org/10.1063/1.3683220]


I. INTRODUCTION
The physical and chemical properties of the rare earth element ytterbium (Yb) show many parallels to those of the alkaline earth elements.For example, Yb is a soft and malleable metal with a relatively low first ionization energy.Furthermore, while its prime oxidation state is +3, Yb also has a stable +2 oxidation state in many chemical compounds. 1his similarity with the alkaline earth metals derives from the filled and compact 4f sub-shell of Yb, which leaves it with a residual valence configuration of 6s 2 .Also like the alkaline earths, as well as the alkali metals, Yb dissolves in liquid ammonia. 2,3 hen alkali metals dissolve in liquid ammonia, solvated electrons form. 4However, when alkaline earth and rare earth metals dissolve in liquid ammonia they do not appear to produce solvated electrons, an observation that can be attributed to their higher ionization energies when compared to the alkali metals.Thus, it is energetically more demanding to persuade an electron to depart from the metal and enter the solvent medium.Nevertheless, such metals could yield solvated electrons if they were transferred into an electronic excited state.
One way of extracting information on the microscopic behaviour of metal atoms in solvents such as ammonia is through the study of metal-solvent complexes in the gas phase.In the case of the alkali metals their complexes with NH 3 have received considerable interest in the past two decades.The earliest experimental work focused on photoionization, which was used (among other things) to extract indirect information on solvent shell closings. 5,6  more recent and state-of-the-art photoionization study has seen the photoionization data extrapolated to exceptionally large complexes with as many as 1500 NH 3 molecules attached to a single Na atom. 7Spectroscopic work appeared subsequent to the early photoionization studies, with NaNH 3 being characterized by resonance-enhanced multiphoton ionization (REMPI) a) Author to whom correspondence should be addressed.Electronic mail: andrew.ellis@le.ac.uk.Tel.: +44 (0)116 252 2138.Fax: +44 (0)116 252 3789.
spectroscopy. 8,9 EMPI is unsuitable for larger Na(NH 3 ) n complexes because of rapid excited state decay and so spectra of these complexes, with up to n = 22, were obtained in the near-infrared region using a photodepletion technique. 10The spectra obtained were attributed to perturbed 3p ← 3s transitions on the Na atoms and these shift strongly to the red in moving from n = 1 to n = 4, with very little change thereafter.
In our own laboratory we have exploited the photodepletion technique to obtain the first mid-IR spectra of Li(NH 3 ) n and Na(NH 3 ) n in the N-H stretching regions. 11,12 hese spectra have allowed a definitive identification of the first solvent shell closings.More recently, we have turned our attention to electronic spectra of alkali-ammonia complexes with the recording of the first electronic spectra of LiNH 3 and Li(NH 3 ) 4 . 13,14 or the latter we found a spectrum in the nearinfrared which has much in common with the spectrum of an electron in liquid ammonia.Very recently, we have extended this work to include the alkaline earth metals and will report on this elsewhere.
Here, we present the findings from our first attempt to record gas phase spectra of a rare earth metal-ammonia complex, where the metal is Yb.We focus on the simplest species, YbNH 3 and its deuterated analogue, YbND 3 .While there have been many spectroscopic studies of Yb-containing diatomics, most notably YbF and YbO, 15,16 there has been only one previous study of a polyatomic Yb-containing molecule, YbOH. 17In the current work we have successfully recorded the electronic spectrum of YbNH 3 using two-color, two-photon resonance-enhanced multiphoton ionization spectroscopy.As will be described, the spectrum obtained is assigned to a spin-forbidden metal-centered transition on the Yb atom.

II. EXPERIMENTAL
2][13][14] Briefly, Yb(NH 3 ) n clusters were produced by laser ablation of a solid ytterbium target in the presence of gaseous NH 3 and the resulting mixture was expanded into vacuum to form a supersonic jet.The central portion of the jet was extracted by a skimmer (2 mm aperture diameter) and the resulting molecular beam passed into the source region of a time-of-flight mass spectrometer, where it was photoionized by the output from a frequency-doubled pulsed dye laser pumped by a Nd:YAG laser and operating using Rhodamine 6G dye.A range of Yb(NH 3 ) n + cluster ions, produced by non-resonant two-photon ionization, could be observed at sufficiently high laser fluence, but the focus here is on the n = 1 cluster.
To record optical spectra, which was achieved using twocolor REMPI spectroscopy, a second laser beam was added to electronically excite the neutral YbNH 3 complex.A LaserVision optical parametric oscillator/amplifier (OPO/A) was employed for this purpose.This system was pumped by the output from an injection-seeded Nd:YAG laser (Surelite II-10), although the injection seeder was not essential for the experiments described here since it did not affect the observable resolution.Since all of the spectra were recorded in the nearinfrared, only the OPO part of the OPO/A system was used.This gave a wavelength tunable output with a pulse energy of up to 15 mJ, although only a small fraction of this available energy was used in the current study.The linewidth of the OPO output with the injection seeder switched off is expected to be ∼3 cm −1 .The OPO and photoionization laser beams were aligned into the source region of the mass spectrometer in a counter-propagating manner in order to achieve maximum spatial overlap between the two beams.The photoionization laser intensity was then reduced to minimize background YbNH 3 + produced by non-resonant two-photon ionization.The temporal delay between the firing of the two laser pulses was controlled using a delay generator to optimize the observed REMPI signal when the OPO wavelength was tuned to an electronic transition of YbNH 3 .The wavelength of the OPO was calibrated using a Burleigh wavemeter.
Ammonia was sourced from a standard liquid ammonia cylinder and was used with no further purification.To assist with spectra assignments some experiments were also carried out using ND 3 (Sigma Aldrich, 99% D-atom substitution).

III. COMPUTATIONAL DETAILS
Ab initio geometry optimization and harmonic vibrational frequency calculations were carried out on YbNH 3 to assist with the spectroscopic analysis.These calculations employed complete active space self-consistent field (CASSCF) or MP2 methodology (see below) and were run using the GAUSSIAN 03 suite of programs. 18For N and H atoms, the standard Dunning cc-pVDZ basis sets were used. 19However, Yb is not included in the Dunning basis sets, and furthermore the large number of core electrons in this atom requires the use of a pseudopotential.Consequently, a Yb basis set and pseudopotential were downloaded from the EMSL basis set exchange website. 20This basis set was of double-zeta quality and was designed for use with a pseudopotential: in this case the Stuttgart ECP28MWB pseudopotential was employed and the combined basis set/pseudopotential package is known as Stuttgart RSC 1997 ECP.
Calculations were begun at the MP2 level with separate calculations on the singlet and triplet manifolds.The expectation was that the ground states in both the singlet and triplet manifolds would correspond to the two electronic states assigned in the spectroscopic work (see later).However, although the triplet state MP2 calculations resulted in optimized geometries with no imaginary vibrational frequencies, the Yb virtual orbitals continually switched order (despite a number of attempts to force the orbitals into the correct order) giving a physically unrealistic situation in which the Yb 7s orbital was below the 6p orbitals, and thus giving a triplet state of the wrong spatial symmetry.Fortunately, the CASSCF calculations successfully bypassed this problem.The active space comprised two electrons and eight molecular orbitals.The specific orbitals in the active space were derived from the combination of the Yb 6s and 6p orbitals with what essentially amounts to the N 2s and 2p valence orbitals.This gives two a 1 orbitals and one e bonding orbital and their antibonding counterparts.The CASSCF calculations were performed to investigate the lowest energy singlet and triplet electronic states, in addition to the first singlet excited electronic state.

A. Photoionization mass spectrometry
Figure 1(a) shows the photoionization mass spectrum derived from the Yb/NH 3 system with the OPO laser beam off-resonance with any spectral transitions of YbNH 3 .Many peaks can be seen, all of which can be assigned to Yb + and various [Yb(NH 3 ) n ] + complexes extending up to n = 16.At the photoionization wavelength used in the present work, 282 nm, there are signal maxima corresponding to n = 7 and n = 13 but we are cautious in attaching any significance to these observations because the ion cluster distribution was found to depend on several factors, including the UV laser wavelength, the laser fluence, and the supersonic expansion conditions.We are planning to explore this photoionization behaviour in detail, including the determination of ion appearance energies as a function of cluster size, and will report on this elsewhere.
The focus of this work is on YbNH 3 , and in particular the dramatic enhancement in the YbNH 3 + signal when the OPO output was tuned to an appropriate wavelength.The mass spectrum shown in Figure 1(b) shows the effects of resonant absorption of the IR OPO beam.A large increase in the YbNH 3 + signal occurs, such that the YbNH 3 + signal is now far larger than the Yb + signal, and this is attributed to REMPI excitation of YbNH 3 .By measuring the YbNH 3 + ion signal as a function of OPO wavelength a two-color REMPI spectrum of YbNH 3 was obtained.
The observation of resonance-enhanced increases in ion signal in the YbNH 3 + ion channel does not rule out any contribution to the REMPI spectrum from larger Yb(NH 3 ) n clusters, but we think contributions from optical excitation of n ≥ 2 clusters are highly unlikely for several reasons.First, we observe no ion signal enhancements in any higher mass channels in the wavelength range investigated, which means that if n ≥ 2 clusters did contribute, then 100% of the resulting ions must fragment rapidly to yield YbNH 3 + .Second, as will be explained in Secs.IV B and IV C, the structure observed in the REMPI spectra is consistent with our expectations for YbNH 3 , and in particular a species with high point group symmetry.Finally, we draw an analogy with Li(NH 3 ) n and Na(NH 3 ) n clusters, where the optical spectra for the lowest lying electronic transitions for n ≥ 2 are strongly redshifted from the n = 1 species. 10,13 urthermore, Li(NH 3 ) n and Na(NH 3 ) n clusters undergo rapid non-radiative relaxation in the excited electronic state, rendering them all unobservable by REMPI except for n = 1. 8,9,14 Athough there is no precise correlation between the electronic states of Li(NH 3 ) n /Na(NH 3 ) n and Yb(NH 3 ) n , as detailed below the strongly metal-centered nature of their low-lying electronic transitions would lead us to expect broadly similar behaviour.For all of these reasons we can confidently assign the spectrum in Figure 2 to YbNH 3 .ferences in the optical spectra were found to be resolvable when choosing isotopologues arising from different Yb isotopes.The lowest energy band in the spectrum of YbNH 3 is observed at 13 795 cm −1 .After substitution of all H-atoms with D, the equivalent band is redshifted to 13 770 cm −1 .There are no bands to the red of this feature, and it therefore seems reasonable to infer that this band represents an electronic band origin.

B. Electronic structure
There are several bands to higher energy in the spectra of both YbNH 3 and YbND 3 .Some of these bands derive from vibrational structure.Indeed, at first sight there looks to be a very obvious low frequency vibrational progression in the REMPI spectra.However, the situation is more complex.Considering the third lowest energy feature in the spectra of both YbNH 3 and YbND 3 for illustration, close inspection shows that there are at least two bands here, with a shoulder appearing on the low energy side of a stronger band.For reasons that will become clear as the discussion progresses, we assign the stronger feature to another electronic origin transition, rather than to vibrational structure.We defer discussion of any vibrational structure to Sec.IV C.
As shown in Table I, bands in the spectrum of YbNH 3 at 13 795, 14 282, and 14 755 are assigned to spin-orbit sub-states in the excited electronic state, specifically to the a The excited state spin-orbit component is identified in parentheses.
3 E 0 , 3 E 1 , and 3 E 2 components, respectively.The justification for this is as follows.First, we refer to the ab initio calculations which reveal that in the ground electronic state the YbNH 3 molecule adopts C 3v point group symmetry.The Yb atom binds to the N atom and the resulting electronic state has 1 A 1 symmetry.Given the availability of low-lying excited states of atomic Yb, which by contrast are not available for NH 3 , we expect the lowest energy electronic transitions of YbNH 3 to be largely metal centered.The same behaviour is seen for LiNH 3 and NaNH 3 for exactly the same reasons. 8,9,14 Wcan therefore draw upon the known electronic energy levels of Yb to construct a simple electronic structure model for YbNH 3 .
The lowest energy electronic transition in atomic Yb involves excitation of one of the 6s electrons to a 6p orbital. 21his excitation can occur as both spin conserved (singletsinglet) and spin-forbidden (singlet-triplet) and, because of its relatively high nuclear charge, spin-forbidden transitions are significant for Yb.In a metal-centered picture of YbNH 3 , the 6s6p 1 P excited state of Yb will correlate with 1 E and 1 A 1 states of YbNH 3 .Likewise, the 3 P excited state of Yb will correlate with 3 E and 3 A 1 states of YbNH 3 .However, only the 3 E state of YbNH 3 can show any spin-orbit structure, and thus we provisionally assign the spectra in Figure 2 to three spin-orbit components of the lowest lying 3 E electronic states of YbNH 3 (upper panel) and YbND 3 (lower panel), i.e., to the ã 3 E − X 1 A 1 electronic system.
The lowest energy electronic transitions identified for LiNH 3 and NaNH 3 , which are largely 2p ← 2s and 3p ← 3s in character, respectively, show redshifts of ∼3500 and 4700 cm −1 relative to the corresponding bare atom transitions. 8,14 y analogy with these complexes we expect a significant redshift for the YbNH 3 6p ← 6s transitions relative to the bare atomic Yb transitions.In atomic Yb the lowest energy spectroscopic transition from the ground electronic state is spin-forbidden.This 6s6p 3 P 0 -6s 2 1 S atomic transition is observed at 17 288 cm −1 and thus, with the expected redshift taken into account, the observed spectral features for YbNH 3 and YbND 3 are located in a region where it is plausible to assign the bands to the ã 3 E − X 1 A 1 electronic transition.By comparison the lowest lying spin-allowed transi-tion of atomic Yb, the 6s6p 1 P-6s 2 1 S transition, occurs at 25 026 cm −1 . 21An exceptionally large redshift would be required if the transition responsible for the spectra in Figure 2 were to correlate with the 1 P-1 S transition.Such a large shift would require very strong perturbation of the unpaired electron density by the NH 3 group and there is no evidence for such an effect.In any case, we can rule out such a transition through the observation of excited state spin-orbit structure.
We can also use the ab initio calculations to extract information on the various electronic states of YbNH 3 , although our primary aim in performing these calculations was to obtain information on geometries and vibrational frequencies, since electronic transition frequencies are difficult to predict without a good quality treatment of the dynamic electron correlation.The CASSCF calculations predict that the ã 3 E − X 1 A 1 and Ã1 E − X 1 A 1 transitions will occur at 22 519 and 29 409 cm −1 , respectively, which puts them well above the experimental region explored in this paper.Nevertheless, the ã 3 E − X 1 A 1 transition is predicted to lie well below the Ã 1 E − X 1 A 1 transition.We can benchmark our calculations by performing corresponding CASSCF calculations on atomic Yb.These predict 6s6p 3 P-6s 2 1 S and 6s6p 1 P-6s 2 1 S transitions at 28 122 and 32 866 cm −1 , respectively, which again are considerably above the experimental values.Consequently, while the CASSCF calculations do a poor job at predicting the absolute transition wavenumbers for the lowest singlet-singlet and singlet-triplet transitions of both YbNH 3 and atomic Yb, they do predict a redshift on addition of NH 3 and they also predict that the singlet-triplet transition will lie well below the singlet-singlet transition.
We now turn to the magnitude of the excited state spinorbit splitting.For a linear molecule in the Russell-Saunders limit, each spin-orbit component should be located at position T 0 + A , where T 0 represents the unperturbed electronic transition frequency (i.e., in the absence of spin-orbit coupling), is the projection of the electronic orbital angular momentum on the C 3 symmetry axis in the excited state, is the corresponding value for the spin angular momentum, and A is the spin-orbit coupling constant.Russell-Saunders coupling will only approximately apply to YbNH 3 because of the heavy Yb atom, and of course it is a nonlinear molecule and so in particular is no longer a good quantum number.The 3 P state of atomic Yb splits into three spin-orbit components, 3 P 0 , 3 P 1 , and 3 P 2 , which are observed at 17 288, 17 992, and 19 710 cm −1 , respectively, relative to the 1 S ground electronic state. 20This implies a spin-orbit coupling constant, ζ 6p , of 704 cm −1 for atomic Yb.In order to gain an estimate of A in the excited state of YbNH 3 , a simple calculation based on the ab initio molecular orbital simulations was performed, i.e., where c i are the coefficients for those atomic orbitals making a substantial contribution to the excited state MO containing the unpaired electron and ζ i is the spin-orbit coupling constant for atomic orbital i. 22 According to our ab initio calculations, the unpaired electron in the ã 3 E state of YbNH 3 is in an orbital strongly dominated by the Yb 6p x,y orbitals, with a much smaller contribution from the N 2p x,y orbitals.The respective normalized coefficients were 0.954 and 0.300, and thus inserting these into Eq.( 1) with the appropriate atomic orbital spin-orbit coupling coefficients leads to a predicted spin-orbit coupling constant, A YbNH 3 , of ∼644 cm −1 .However, the off-axis H atoms in the YbNH 3 complex are expected to partially quench the electronic orbital-angular momentum and, therefore, reduce the effective spin-orbit coupling.Consequently, 644 cm −1 should be an upper limit to the true spinorbit coupling constant of the ã 3 E state.
The 3 E 1 -3 E 0 and 3 E 2 -3 E 1 separations observed from the YbNH 3 spectra are 487 and 473 cm −1 , which are consistent with the theoretically derived value when the partial quenching of the electronic orbital angular momentum is taken into account.It is noteworthy that A YbNH 3 = A YbND 3 with the latter reduced to ∼440 cm −1 .This difference is not huge, but it is significant given that the spin-orbit coupling would normally be expected to be isotope invariant.The fact that this is not, suggests that substantial perturbations are in action, as noted recently for the Ã 2 states of BaOH and BaOD. 23The most likely candidate for the perturbing state is the lowest 3 A 1 state, which like the ã 3 E state correlates with the 6s6p 3 P excited state.We have not located the 3 A 1 state but it is likely to be reasonably close to the ã 3 E state and must have a differential perturbing effect on the two isotopologues, YbNH 3 and YbND 3 , in order to generate different effective spin-orbit coupling constants.

C. Vibrational structure
We now turn to the remaining features in the REMPI spectra of YbNH 3 and YbND 3 , which are attributed to vibrational structure.YbNH 3 will have six vibrational normal modes, with approximate descriptions and symmetries as follows: the Yb-N stretch (a 1 ), the Yb-N-H bend (e), the NH 3 umbrella vibration (a 1 ), the H-N-H bending mode (e), the symmetric N-H stretch (a 1 ), and the antisymmetric N-H stretch (e).To gain some idea of the frequencies of these vibrations, we show the ab initio predictions for YbNH 3 and YbND 3 in several electronic states in Table II.
Most of the vibrational structure seen in the YbNH 3 spectrum can be accounted for by excitation of the Yb-N stretching vibration, ν 3 .The first band to the blue of the ã 3 E0 − X 1 A 1 0 0 0 band is a strong band located at +232 cm relative to the origin band.This interval is very close to the ab initio prediction for the Yb-N stretching vibrational frequency in the ã 3 E state.Other members of the ν 3 progression associated with this spin-orbit component, as well as the other spin-orbit components, can be readily found, as shown in both Figure 2 and as summarized in Table I.These assignments are supported by the small shifts in the measured vibrational intervals on deuteration, which would be inconsistent with all vibrations except the Li-N stretch.For example, we measure a 3 1 -3 0 interval of 217 cm −1 for YbND 3 , which is 15 cm −1 smaller than that determined for YbNH 3 .This compares very well with the redshift of 17 cm −1 for the ã 3 E state predicted from the ab initio calculations (see Table II).Furthermore, we note that a substantial reduction in the Yb-N equilibrium bond length, from 2.72 to 2.59 Å, is predicted by the CASSCF calculations for excitation from the X 1 A 1 state to the ã 3 E state.This is consistent with the observation of a significant progression in this vibrational mode.
The combination of excited state spin-orbit coupling and vibrational structure in mode ν 3 accounts for all of the bands with significant signal-to-noise ratios in Figure 2.There is no evidence for excitation of the Yb-N-H bend (ν 6 ), whose predicted frequency is sufficiently low to be observable in the scan range covered in Figure 2. Structure in this non-totally symmetric vibration would require vibronic coupling to gain significant intensity, but such coupling has been seen previously in spectra of LiNH 3 and NaNH 3 . 8,9,11,12 Hoever, this vibration would be characterized by relatively large isotope shifts on deuteration, which would be inconsistent with the YbNH 3 and YbND 3 spectra in Figure 2. Equally, there is no evidence for excitation of the umbrella mode, although in the case of YbNH 3 this is likely to fall outside of the scan range reported here.The scan range in the current study is constrained by the OPO, whose output pulse energy declines dramatically beyond 15 000 cm −1 .It would, therefore, be interesting to extend this work using dye laser excitation to try and access higher lying vibrational levels in the ã 3 E state, as well as for finding the 3 A 1 state and the singlet electronic excited states.

V. CONCLUSIONS
The first spectroscopic observation of a rare earth metalammonia complex in the gas phase has been reported.The spectrum of YbNH 3 was obtained using two-color REMPI spectroscopy and the features observed have all been assigned to the spin-forbidden ã 3 E − X 1 A 1 electronic band system, which correlates with the lowest 6s6p 3 P ← 6s 2 1 S transition of atomic Yb.The measured spin-orbit coupling constant is close to that expected for the ã 3 E state on the basis of the calculated molecular orbital composition, although it is reduced by off-axis quenching of the orbital angular momentum by the H atoms. Vibrational structure is seen which can be accounted for by excitation of the Yb-N stretching vibration.

FIG. 1 .
FIG. 1. Photoionization mass spectra taken at a UV wavelength of 282 nm and with (a) the OPO laser off-resonance (13 901 cm −1 ) and (b) with the OPO laser on-resonance and exciting the electronic band origin in YbNH 3 (13 795 cm −1 ).The inset to (b) shows the region between m/z 165 and 202, highlighting the resolution obtained in the mass spectrometer; Yb has several observable isotopes and each is clearly observed.Note that plots (a) and (b) do not have the same y-axis scale; in both plots the signal intensity of Yb + is identical.

Figure 2
Figure 2 compares the REMPI spectra of YbNH 3 and YbND 3 .These spectra were recorded by summing the mass spectral signals over all possible Yb isotopes, since no dif-

TABLE I .
Band positions and assignments for the YbNH 3 REMPI spectrum (in cm −1 ) and quoted relative to the YbNH 3 and YbND 3 band origins at 13 795 and 13 770 cm −1 , respectively.

TABLE II .
Ab initio vibrational frequencies of YbNH 3 and YbND 3 (in cm −1 ) calculated at the CASSCF/cc-pVDZ level of theory.