photoionization spectroscopy of auar

ELSEVIER

18 July 1997

Chemical Physics Letters 273 (1997) 265-271

CHEMICAL PHYSICS LETTERS

Photoionization spectroscopy of Au-Ar

A.M. Knight, A. Stangassinger, M.A. Duncan * Department of Chemistry, University of Georgia, Athens, GA 30602, USA

Received 11 March 1997; in final form 8 May 1997

Abstract

Resonant photoionization spectroscopy probes the new metal van der Waals complex Au-Ar. Band systems observed near 270 nm and 245 nm correlate to the spin-orbit components (2PI/2 ~ 2S and 2P3/2 ~ 2S) of the Au (6p ~ 6s) atomic transition. Extrapolated vibronic progressions yield the excited state convergence energy, which is combined with the atomic asymptote to yield the ground state bond energy, D~ = 130 + 15 cm- t. Lower limits for the excited state binding energies are Db(2IIt/2 ) >/338 cm - l and Db(21-13/2 ) ~ 654 cm -I . The ionization potential is ~< 9.208 eV. Au-Ar is compared to the Cu-RG and Ag-RG complexes studied previously. © 1997 Elsevier Science B.V.

1. Introduct ion

Much work has been done in recent years to probe the weak bonding interactions exemplified by van der Waals complexes of rare gas (RG) atoms with metal atoms [1-4]. Increased understanding of this fundamental bonding provides improved models for many chemical systems of interest, in particular physisorption on metal surfaces and metal- l igand bonding. The meta l -RG complexes have electronic absorptions near the atomic resonances on the metal atom. Of particular interest are the noble metal atoms which have an alkali-like valence electronic structure, ndl°(n + 1)s 1. In these systems, the strong absorption of the atomic resonance 2p ~ 2 S on the metal atom gives rise to corresponding strong molecular absorptions in the same wavelength region for me ta l -RG complexes. Studies have been reported previously by our laboratory for A g - R G [2] and for C u - K r [3] but the present example of A u - A r pro-

* Corresponding author.

vides the first data for a gold complex. The comparison of bond energies across the noble metals provides new insights into the details of the metal van der Waals interaction.

The previous work on noble metal-rare gas complexes, including the systematic comparison of different metals and rare gases, has revealed several g e n e r a l c h a r a c t e r i s t i c s o f m e t a l - l i g a n d interactions[I-4]. In each of these isoelectronic systems, the ground electronic state resulting from the M(2S) + Ar( IS) combination is z]~, while both 21I and 2~ excited states correlate with the M ( 2 p ) + Ar (l S) asymptote. The electrostatic bonding is quite different in these electronic states due to the specific orbital configurations and their influence on the van der Waals interaction. As a result of these effects, the excited state binding energy, Db( 2 II) , is found to be consistently greater than that of the corresponding ground state, D~( 2 Z), in each of the noble meta l -RG systems. For a given metal atom, complexes with heavier rare gases are found to be more strongly bound, i.e., Do(M-Xe) > Do(M-Kr) > Do(At), as expected from the increased polarizability of the

0009-2614/97/$17.00 © 1997 Elsevier Science B.V. All rights reserved. PII S0009-2614(97)00568-X

266 A.M. Knight et al. / Chemical Physics Letters 273 (1997) 265-271

heavier rare gases. However, systematic comparison of different metals produces binding energy trends that cannot be explained with simple polarizability arguments. For example, the binding energy for C u - Kr was found to be greater than that for A g - K r [2,3], even though the polarizability of silver atom is greater than that of copper (O~cu = 6. l × 10 -24 cm3: aA~ = 7.2 X 10 -24 cm3)[5]. This observation was rational- ized as arising from a valence orbital configuration interaction. The copper atom has a low-lying 2D atomic state corresponding to the 3d94s 2 electron configuration[6], and the blending of d orbital character into the molecular ground state makes it possible to reduce valence shell electron repulsion, thus strengthening the overall bonding. This additional orbital effect could also exist for the gold atom because it has a low lying 2D atomic state at approx- imately the same energy as that for copper[6]. The gold atom has a lower polarizability than that of either silver or copper (aAu = 5.8 X 10 -24 cm3)[5], and so simple polarizability trends would predict that its rare gas complexes would be more weakly bound. However, as shown below, the A u - A r complex is more strongly bound than Ag-Ar , providing additional support for d orbital configuration interaction in the bonding of these complexes.

2. Experimental

The A u - A r complex is made by laser vaporiza- tion in a pulsed nozzle cluster source using a 1 / 4 inch diameter solid gold rod and a pure Ar expan- sion. A XeC1 excimer laser (Lumonics) at 308 nm is used to vaporize the sample. The gas pulse is intro- duced using a modified Newport BV-100 valve [7]. The resulting molecular beam then passes into the detection chamber where the complex is photoion- ized and the resulting ions are mass analyzed with a linear time-of-flight spectrometer. The molecular beam apparatus has been described previously [8]. The one-color two-photon photoionization scheme employs a Nd:YAG-pumped dye laser (Spectra- Physics DCR-3 /Lumonics HD-500). Part of the data are collected using a manually-tuned KDP crystal for frequency-doubling, while the rest are collected using the Lumonics HypoTrack-1000 auto-tuner. The signal is monitored in both the parent molecular ion

channel (237 amu) and the metal atomic ion channel (197 amu). The spectra are calibrated using known atomic Au resonances.

3. Results and discussion

Two band systems are observed for A u - A r (see Fig. 1) near 270 nm and 245 nm. These molecular transitions must correlate to the two spin-orbit components (2pl/2 ~ 2S and 2p3/2 *-- 2S at 37358.9 and 41174.3 cm -1 respectively) of the Au ( 6 p , - 6 s ) atomic transition because there are no other atomic states in this energy region [6]. The position of the respective atomic lines are shown as dotted lines in each spectrum. Lower lying A, B and C excited states are expected which would correlate to the Au(2D) + Ar asymptote. The spectra observed are therefore assigned to the D 2 I ~ l / 2 ~ - - X 2 ~ and D 2H3/2 ,,-- X 2~ transitions, respectively, as indi- cated in the figure. While we recognize that the large spin-orbit interaction in Au makes it more appropriate to describe these systems in Hund's Case c, we

¢v

21-[ 1 / 2 ~ - 2Z

| - - i ~ , 37100

A u A r

I 3 7 2 0 0 3 7 3 0 0 3 ? 4 0 0 3 7 5 0 0

E n e r g y ( c m - l )

e-,

i I q I

4 0 5 0 0 4 0 7 0 0 4 0 9 0 0 41 I 0 0 E n e r g y ( c m - t )

Fig. 1. The photoionization electronic spectra of Au+-Ar. The two band systems observed correlate to the two spin-orbit components of the Au (2Pi/2,3/2 ,-- 2S) atomic transition. The dotted vertical lines indicate the positions of the atomic transitions.

A•M. Knight et al. / Chemical Physics Letters 273 (1997) 265-271 267

Tab le 1

B a n d posi t ions o f the 2111/2 <--2]g t ransi t ion o f A u - A r (t ," = 0)

L" E n e r g y ( c m - 1 )

n 37135.1

n + 1 37200 .7

n + 2 37258 .9

n + 3 37310.1

n + 4 37353.2

n + 5 37389.1

n + 6 37415 .6

n + 7 37436 .5

n + 8 37450 .8

n + 9 37463 .2

n + 10 37473.7

use the Case a notation for convenient comparisons to other corresponding systems. The spectra shown are recorded in the parent ion Au +-Ar mass channel, but the same spectrum is also measured in the Au + mass channel for the D 2113/2 ~ X 2~ system. Pro- gressions with clear anharmonicity extending for 11 and 13 bands are observed in these systems. The line positions for these bands are given in Tables 1 and 2. There is an underlying hump in the higher energy region of each spectrum, where broad continuous signal underlies the sharp lines.

The appearance of the bands in both systems is consistent with a vibronic progression from v" = 0 in the ground state to a series of levels in the excited state. In the D2II1/2 ~-X 2E system, the spacings between progression members in the excited state begin at about 65 cm -1 and decrease to only l0 c m - 1 at the highest energy bands. In the D 2II3/2 ~-

Tab le 2

B a n d pos i t ions o f the 2133/2 ,-2~ t ransi t ion o f A u - A r (c" = O)

t~' E n e r g y ( c m - 1 )

n 40481 .2

n + 1 40556 .7

n + 2 40629 .0

n + 3 40696 .4

n + 4 40760 .6

n + 5 40819 .8

n + 6 40876 .3

n + 7 40928 4

n + 8 40977 .3 n + 9 41020 .3

n + 10 41063 .0

n + l l 41100 .3

n + 12 41135 .3

X 2 E system, the spacings between progression members in the excited state begin at about 76 cm- and decrease to 35 cm -~ at the highest energy bands. Unfortunately, it is not possible to assign these progressions to specific quantum number transitions. Au-Ar exists as only one naturally occurring isotopomer which has a mass of 237 amu, and therefore isotopic shift analysis cannot be performed to confirm the assignments• Because of the gradual onset in intensity in both band systems, and the long progressions observed, it is very unlikely that the first lines observed correspond to the electronic ori- gins. However, these progressions can be fit reason- ably well with an equation of the form E(t~)= a(t~ + l / 2 ) - ¢ot x'~(u + 1/2) 2, where a must be taken as a lower limit to to'~. This analysis provides the constants a -- 73.3, tOeX' e = 3.77 for the 21-I./2 state (using only the first 9 transitions; see below) and a = 79.4, t%x'~ = 1.93 for the 2113/2 state.

Even though a complete vibronic analysis of these spectra cannot be obtained, several conclusions can be reached about the nature of the van der Waals bonding. First of all, the observation of structured spectra beginning to the red of the corresponding atomic transition immediately establishes that the

• 2 excited l-I1/23/2 states are both more strongly bound 2 ' • • • than the E ground state. A similar observation was

reported previously for the corresponding Ag-RG and Cu-Kr systems[2,3], where the 21I excited states

• . 2 correlating to the metal atomic Pj/23/2 state are L / • also more strongly bound than the ~ ground state.

In these 2II excited states, the valence electron occupies a p~ orbital with most of the electron density off the intermolecular axis, which results in reduced valence shell electron-electron repulsion. The positive core of the metal atom is also exposed to the rare gas atom, enhancing the polarization which provides the attractive force for the system. The lower repulsion coupled with increased attrac- tion makes the excited-state relatively strongly bound. In contrast, the s(r-type interaction keeps the electron density on-axis and there is significant repulsion between the metal and rare gas valence shells, which produces a relatively weak ground state 2~ bond. Secondly, the range of bound vibrational levels observed provides a lower limit on the well depth in the excited states. In the 2H1/2 state, eleven bands are observed whose energy span provides the limit


D~(2II1/2) >~ 338 cm -1. In the 21][3/2 state, 13 bands are observed whose energy span provides the limit D~(2II3/2) >~ 654 cm - j . Finally, it is worth noting that the Franck-Condon access to the 2IIj/2 and 21-I3/2 excited states is at very different positions relative to their dissociation limits. This suggests that these excited states have very different potentials and bonding, even though they correspond to the same atomic orbital configuration. This is perhaps not too surprising for spin-orbit components spaced so widely apart in energy (3815 cm-1 atomic spin-orbit splitting).

A more detailed investigation of the van der Waals bond energies in this system is possible by an extrapolation of the vibronic bands observed to the dissociation limit. The anharmonicity is clearly significant in both band spectra, but it is quite extreme in the 2I-Ii/2 ,,--X 2]~ system. The last few levels deviate severely from the positions predicted by the two-parameter vibrational energy equation, and so these levels were not included in the fit to determine the constants noted above. There are two commonly used methods for the extrapolation of vibronic levels to determine dissociation energies. In the Birge- Sponer analysis, the decrease in vibrational spacing as the limit is approached is assumed to be linear[9], while in the Le Roy-Bernstein analysis the trend is described by a power law, with an appropriate expo- nent to describe the exact electrostatic interactions present[10,11]. The Le Roy-Bernstein methodology is especially appropriate if levels at the convergence limit are measured, while the Birge-Sponer method, which is equivalent to the assumption of a Morse potential, may not be adequate for weakly interacting systems near the dissociation limit. We have applied both methods here to investigate the dissociation energetics and to explore the applicability of these methods.

Simple inspection of the line spacings in the 2111/2 ~ X 2~ indicates that the levels measured are closer to the dissociation limit than those in the 2I/3/2 ~ X 2 ~ system. A L e Roy-Bernstein extrapolation is shown in Fig. 2. It plots the vibrational spacing to the 3 / 2 power as appropriate for the 1/r 6 attractive interaction, and produces a convergence limit of 37487 cm -1, which is 14 cm -1 above the last band measured. A Birge-Sponer plot in its normal form requires the knowledge of quantum

_,..,,

t'-q

L3

500 •

400

300-

200-

100"

0 37185.0

Au-Ar Leroy-Bernstein Plot

2H1/2-~ 22

0 0

37263.8 37342.5 37421,3 37500.0

Energy (cm -1) Fig. 2. The Le Roy-Bernstein extrapolation of the vibrational levels in the 2IIi/2 , , -X 2~ band system. The convergence limit determined from this data is used to derive the dissociation energy.

numbers, which is not available here. However, a simple plot of vibrational spacing versus energy is in principle equivalent to the Birge-Sponer plot, and such an extrapolation produces a convergence limit at 37530 cm -1, which is 63 cm - l above the last band measured. The last few line positions deviate significantly from the linear plot, indicating that this analysis overshoots the exact limit. The Le Roy- Bernstein plot is expected to perform best when levels at the dissociation limit are measured. There- fore, it seems that the limit obtained with the Le Roy-Bernstein method represents the most accurate determination of the convergence limit. Similar extrapolations are carried out for the 2I-I3/2 ~ X 2 ~

system, but the extrapolation by both methods is much longer. The Le Roy-Bernstein method produces a limit at 41380 cm -~, which is 245 cm - l beyond the last line observed. The linear extrapolation produces a limit at 41640 cm-1, which is 505 cm-1 beyond the last line. The agreement between the two methods is much worse here, as might be expected for a long extrapolation. Additionally, both of these methods are likely to overshoot the binding when only levels low in the potential are available. Because the levels just at the convergence limit are measured in the 2H1/2 ~ X 2]~ system and not in the 2II3/2 ~ X 2E system, we use the 2II1/2 ~ X 2]~ system numbers in further analyses below.

The extrapolations described and the convergence limit obtained are useful because we know the atomic

A.M. Knight et al. / Chemical Physics Letters 273 (1997) 265-271 269

transition to which these molecular spectra correlate. Subtraction of the atomic transition energy from the convergence limit measured produces a binding energy for the ground electronic state: D 0 = limit(2ii1/2 ,__ 2 ~ ) _ [2P1/2 ~.. 2S] = 37487 -- 37359 = 128 cm -~. Similar treatment of the Le R o y - Bemstein data for the 21-I3/2 *--- X 2]~ system, where the extrapolation is much longer and is likely to over-estimate the true binding, produces a value of D~ = 206 cm -1. Again, the value derived from the 21-/1/2 ~ - - X 2 ~ system is preferred because the extrapolation is much shorter and more likely to be valid. Another experimental fact relevant here is that the spectrum measured for the 2111/2 *--X 2 ~ system has lines appearing above the position of the corresponding atomic transition. The energy of the last line (37474 cm -1) is 115 cm -1 above the atomic transition, which places a rigorous lower limit on the ground state dissociation energy. As a conser- vative estimate of the ground state dissociation energy then, we choose the value 130 + 15 c m - ~.

The polarizabilities of the metal and rare gas atoms are important factors in van der Waals bonding. The values for the noble metals are: aCu = 6.1 X 10 -24 cm3; O~Ag = 7.2 X 10 -24 cm3; and OgAu =

5.8 X 10 -24 cm 3, while the value for argon is ~Ar = 1.62 × 10 -24 cm 3. Thus, the metal atom polarizability is much greater than that of argon, and variations between the values for the different metal atoms should have a noticeable effect on van der Waals bonding. The ground state dissociation energy for Ag-Ar has been determined previously to be 65 + 20 cm-1 [2]. It is therefore apparent that the van der Waals bond energy for A u - A r is greater than that in Ag-Ar, even though Ag is more polarizable than Au. A similar trend was observed previously in the comparison of Cu-Kr (D~ = 408 cm -1 ) and A g - K r (D 0 = 138 cm-1), where the copper complex was more strongly bound even though silver has a greater polarizability [3]. In each of these pairs of complexes with a common rare gas, the metal with the lower polarizability binds more strongly, contrary to the expected trend for 'simple' van der Waals interactions. Apparently, additional details beyond polarizability play a significant role in these weak interactions. The likely explanation, which has been discussed before, involves the detailed nature of the valence metal orbitals. The ground state for all three

metals is 2~, which correlates to M (2p) + Ar (ip). Differences arise in the position of the atomic 2D state on the metal, which corresponds to the ndg(n + 1)s 2 configuration. The Cu atom has a low-lying 2D state (11203/13245 c m - l ) , as does Au (9161/21435 cm-1), whereas the analogous 2D state lies much higher in energy for Ag (29552/30473 cm-1). The low-lying 2D states in Au and Cu allow sd-hybridization to take place in the ground state of metal -RG complexes. Such configuration mixing adds off-axis components into the wavefunction, which in turn leads to a reduction in the M - R G valence-shell repulsion and a consequent increase in van der Waals bonding strength. This effect was suggested initially by Breckenridge [12], and it was used to explain the bonding in Cu-Kr and Ag-Kr complexes [3]. The same kind of interaction is apparently now confirmed again in the pattern of bond strengths seen for Au-Ar versus Ag-Ar. It would of course be interesting to have bond energies for Cu- Ar and Au-Kr to complete these comparisons. How- ever, we have not yet been able to obtain spectra for Cu-Ar from its ground state. Studies of additional gold complexes are now underway. It would also be interesting to have ab initio calculations of these interactions to get a more quantitative picture of the degree of configuration mixing which occurs.

An additional point of interest here is the significant differences apparent for the two excited state potentials. As shown in Fig. 1, the Franck-Condon region of the e x c i t e d 2[I1/2 state extends well beyond the atomic resonance line, accessing vibrational levels extending almost to the dissociation limit. In the 2113/2 state, the Franck-Condon region lies lower in the potential and the observed progression ends before the atomic resonance line. As pointed out to us by Breckenridge [12], such effects are understand- able in light of the strong spin-orbit interaction for Au. The spin-orbit interaction leads to extensive mixing of molecular states with the same values of g2 = 1/2, i.e., the e x c i t e d 2~?/2 state (not observed and believed to be largely repulsive in its bonding) and the excited 2H ~/2 state, which has much stronger electrostatic bonding as discussed above. These molecular states correlate to the Au (5p 2p3/2) and Au (5p 2P~/2) atomic states respectively. The pure electrostatic bonding effects discussed above can be mitigated partially by spin-orbit mixing of these


states. Therefore, the 2111/2 state will be less bound than expected because of the ~ character mixed into the wavefunction by the spin-orbit interaction and

2 - + the £~/2 state will be less repulsive than expected due to the rr character mixed into the wavefunction by the spin-orbit interaction. The 21/3/2 state would be unaffected by this mixing and would have essen- tially pure ar character in its electrostatic bonding. The cr /w mixing is likely to be most significant at large internuclear separations where A~.o. is greater than the electrostatic binding energy, which is ex- actly the region of these potentials accessed in the spectrum. Therefore, in the Franck-Condon regions of the excited 211 states, the curvatures of the potentials are likely to be very different as a result of spin--orbit mixing, and this could easily explain the qualitative differences in the appearance of the spectra.

There is a region of continuous signal underlying the sharp structure in both of these spectra. These features occur in the higher energy region and they have a similar profile in each spin-orbit component, although they occur at different positions relative to the excited potential and with respect to the atomic asymptote. One explanation for such a signal is a curve-crossing in the excited state. The lower lying Au(2D) + Ar asymptote should give rise to three molecular states (2~, 21-i, 2A) ' which all might lie energetically at a position to cross the states studied here. If such a curve crossing is in effect, the most likely candidate is the 2H state, which could couple strongly with these states. However, considering that the main progression members here appear as sharp peaks on top of these continuous regions, a more likely explanation is that larger clusters are fragment- ing into the AuAr + mass channels in these regions.

The lowest observed line (37135 cm -1) in the 2111/2 , --X 2~ system gives an upper limit for the ionization potential of the complex. Since this peak is measured efficiently with two-photon absorption, the IP can be no greater than twice the energy of this photon (IP ~< 9.208 eV). Pyykko [13] has calculated the dissociation energy of the Au+-Ar cation to be 0.510 eV. The ionization potential of the complex should then be

IP(Au-Ar) = IP(Au) - [ D0(Au +-Ar)

- D0(Au-ar)] .

Using our estimate for the ground state binding energy (130 cm -1 = 0.016 eV) in combination with Pyykko's number for the ion complex and the atomic value (IPAu = 9.225 eV), a value of 8.731 eV is obtained for the neutral ground state ionization potential. This is consistent with the upper limit observed (9.208 eV). Related ionization energetics pro- vide an explanation for the ion channels produced in these spectra. In the 21/3/2 ~ X 2~ system, the spectrum appears in both the parent ion channel and the Au + fragment channel, while in the 2H1/2 ~ X 2]~ system it appears only in the parent channel. In the ionization process through the 2II1/2 intermediate state, the total two-photon energy from complex ground state exceeds the molecular IP, but it does not reach the Au++ Ar ionization potential asymptote (determined by D~ + IP(Au)). Parent ion formation is therefore the only process energetically possible. Ionization through the 2113/2 state however, exceeds this asymptote and fragment ion formation becomes energetically possible.

Au-Ar provides the second example of a noble metal van der Waals complex which exhibits a surprising trend in its bond energy. Additional studies of the Au-RG series of complexes (RG = Kr, Xe) are currently underway in our group to explore these trends further. The addition of isotopic data in the cases of Kr and Xe will make a more complete vibrational analysis possible, providing more detailed probes of the excited state potential in these systems. The extension of the Cu-RG series, also currently underway in our group, will permit a full comparison of trends in the noble metal -RG series.

Acknowledgements

This research is supported by the US Department of Energy through grant No. DE-FG02-96ER14658.

References

[1] W.H. Breckenridge, C. Jouvet, B. Soep, in: Advances in Metal and Semiconductor Clusters, Vol. III, M.A. Duncan (Ed.), JAI Press, Greenwich, 1995.

[2] L.R. Brock, M.A. Duncan, J. Chem. Phys. 103 (1995) 9200. [3] L.R. Brock, M.A. Duncan, Chem. Phys. Lett. 247 (1995) 18. [4] A. Stangassinger, I. Mane, V.E. Bondybey, Chem. Phys. 201

(1995) 227.

A.M. Knight et al. / Chemical Physics Letters 273 (1997) 265-271 271

[5] M. Rigby, E.B. Smith, W.A. Wakeham, G.C. Maitland, The Forces Between Molecules, Clarendon Press, Oxford, 1986.

[6] C.E. Moore, Atomic Energy Levels, National Standard Ref- erence Data Series 35, National Bureau of Standards, US GPO, Washington, DC, 1971.

[7] UR. Brock, D.L. Robbins, J.S. Pilgrim, M.A. Duncan, Rev. Sci. Instrum. 67 (1996) 2989.

[8] K. LaiHing, R.G. Wheeler, W.L. Wilson, M.A. Duncan, J. Chem. Phys. 87 (1987) 3401.

[9] G. Herzberg, Molecular Spectra and Molecular Structure I. Spectra of Diatomic Molecules, Van Nostrand Reinhold, New York, 1950.

[10] R.J. Le Roy, R.B Bernstein, J. Chem. Phys. 52 (1970) 3869. [11] R.J. Le Roy, R.B. Bernstein, J. Mol. Spectrosc. 37 (1971)

109. [12] W.H. Breckenridge, personal communication. [13] Pyykko, J. Am. Chem. Soc. 117 (1995) 2067.

photoionization spectroscopy of auar

Documents