High-resolution laser spectroscopy of cyclometalated Rh (III)-thienylpyridine complexes

Chemical Physics 173 (1993) 513-523 North-Holland

High-resolution laser spectroscopy of cyclometalated Rh (III)-thienylpyridine

complexes

Arne ZiIian ’ , Gabriela Frei and Hans U. Giidel * Institut

fir

Anorganische Chemie, Universitiir Bern, Freiestrmse 3, CH-3000 Bern 9, Swttzerland

Received 20 October 1992; in final form 25 March I993

The ~clome~lated rh~ium(III) complexes [Rh(~py)~L]+ with L=bpy, en or [~-Cl~-(thpy)~~]are studied in poly(methylmethac~Iate) (FMMA) or in mixed crystals of the corresponding [Rh(phpy)2L) + complexes. High-resolution luminescence and excitation spectra of the complexes at temperatures around 5 K are presented. The vibrational sidebands are fully resolved and allow us to compare the vibrational structure of the ground and lowest excited state. All thpy- complexes show nearly identical vibrational sidebands. The possibilities to obtain an estimate of excited state geometries are discussed.

/

There has been considerable interest in the spectroscopic properties of complexes with chelating ligands like 2,2’-bipyridine (bpy), o-phenanthroline (phen ) , 2-phenylpyridine ( phpyH ) and 2-thienylpyridine (thpyH) f l-31. Among those the (4d)6 complexes belong to a class of compounds for which it has been shown that they can convert light into chemical [ 41 or electrical [ 5) energy, yet the photophysical properties of the involved triplet excited state are not well underst~d. Usually, the excited electronic states of such complexes are characterized by a single orbital configuration such as x-x* (ligand centered, LC) and d-n* (metal-to-ligand charge transfer, MLCT). Whereas the lowest ‘x-x* states lie above the lowest ‘d-x* states, the corresponding triplet states are expected to have comparable energies and therefore the first excited state might be of mixed character. While it has been difficult to obtain sharp-line optical spectra of the triplet state for complexes like Ru(bpy)$+ and Rh(bpy)j+, we could measure highly resolved spectra of cyclome~lated Rh3+ complexes. From a combination of luminescence line ’ Present address: Department of Chemistry, University of Wisconsin, 1101 University Avenue, Madison, WI 53706, USA. * To whom correspondence should be addressed. 0301-0104/93/$06.00

na~owing (LLN) spectroscopy [5-S 1, single crystal absorption [ 9,10 ] and energy migration studies [ 1I ] we came to the conclusion that the lowest excited state has predominantly 3rt-x* character and that it involves the cyclometalating phenylpyridine or thienylpyridine ligands in the mixed chelate complexes [Rh(phpy)~bpyl* and [Rh(thpy)~bpyl~, respectively. Experiments on the behaviour of the electronic origin in a magnetic field [ 12,13 ] and optically detected magnetic resonance (ODMR) experiments [ 14) support this view and there is evidence that a localized description of the lowest triplet state is useful [ 10 1. The present study is devoted to Rh(XI1) tris chelate complexes containing two cyclometalating thienylpyridine ligands. The third chelating &and is either 2,2’-bipyridine ( [Rh(thpy)zbpy J’) or ethylenediamine ( [ Rh (thpy ) 2en ] + ) or the chloro bridge in the case of the chloro bridged diner ( [ Rh ( thpy)zu-CII-(thpy)zRh] ), These [Rh(thpy),L]+ complexes will be embedded in poly (methylmethacrylate) (PMMA) or in crystals of the corresponding [ Rh(phpy)2L] + complexes. The crystal structure of [ Rh (phpy )zbpy 1PF6 is perfectly ordered and we obtain narrow luminescence and excitation spectra of the [ Rh ( thpy)zbpy] + guest complexes. In the other samples the natural linewidth is one to two orders of magnitude bigger due to inhomogeneous broadening. Luminescence and excitation line nar-

0 1993 Elsevier Sctence Publishers B.V. All rights reserved.

514

A. Zilran et al. /Chemical Physrcs I73 (1993) 513-523

rowing techniques (LLN and ELN, respectively) were therefore used. As a result, highly resolved vibrational sidebands could be obtained.

2. Experimental The

binuclear

complexes

[ Rh ( thpy)2-u-ClZ-

(thpy),Rhl and [Rh(phpy)z-u-CIZ-(phpy)ZRhl were synthesized following literature methods [15,16]. In order to obtain [Rh(thpy)zbpy]+, and [Rh(th-v)2enl+, [Rh(pbyMwyl+ [ Rh (phpy ),en] + the parent Cl-bridged dimers were refluxed in dichloromethane for 1 h with 2,2’-bipyridine or ethylenediamine, respectively. The PF; salts were obtained from the Cl- salts by metathesis with NH,PF,. The products gave satisfactory elemental analyses. The identity of the polycrystalline powders was further checked with 400 MHz ‘H NMR spectra and powder X-ray diffraction. [ Rh( thpy ),L] + doped [ Rh(phpy),L] + powders were obtained by cocrystallisation from a dichloromethane solution. [ Rh(thpy ),bpy ] PF, was also doped in PMMA by evaporation of a 10% w/v solution of PMMA in dichloromethane (complex concentration - 1O-3 M). The samples were cooled in a helium-gas flow tube [ 17 1. Luminescence was excited with lines of an Ar+ laser (Spectra Physics 2045 ) or with a Nd: YAG laser (Quanta Ray DCR3D) pumped dye laser (Lambda Physik FL3002). The bandwidth of both excitation sources was better than 0.2 cm-‘. Dye solutions were made from coumarin 152, coumarin 102 and coumarin 47. The luminescence was dispersed by a Spex 1702 single monochromator equipped with a 1200 grooves/mm grating blazed at 300 nm. Detection was achieved by a RCA 3 1034 photomultiplier and a gated photon counting system (Stanford Research SR-400). The gate was started 5 us after the laser pulse and had a duration of 5 ms. In this way, the effect of scattered laser light in the resonant line narrowing experiments could be minimized. The resolution in the LLN and ELN experiments was largely determined by the chosen monochromator resolution ( - 1 cm-‘). The luminescence spectra are corrected for instrumental response. For the excitation spectrum of [ Rh( thpy ),bpy ] + in [Rh(phpy),bpy]PF, light of a 100 W halogen

lamp dispersed by two gratings ( 1200 grooves/mm) blazed at 500 nm of a Spex 1402 double monochromator was used. The resulting luminescence was detected with a RCA 31034 photomultiplier below 17500 cm-’ using a cut-off filter. This spectrum was corrected for instrumental response. The Raman spectrum was excited with the 647 nm line of a Kr+ laser (Coherent CR-500 K) and detected using the same monochromator. The accuracy in the energy determination is estimated to be + 2 cm-’ when the monochromator was scanned and slightly better when the dye laser was scanned. A Tektronix 4052A microcomputer was used for monochromator control and data acquisition.

3. Results and discussion 3.1. Ground state vibrational structure The ground state vibrational structure of [ Rh (thpy ),L] + complexes is accessible both through their Raman and through their luminescence spectra. However, in order to determine the vibrational energies in the electronic ground state, highly resolved spectra are necessary. Fig. 1 shows a survey of luminescence spectra of crystalline samples containing mixed cyclometalated rhodium(II1) complexes. The 5 K luminescence spectra were excited by UV light of an Ar+ laser (at 363.8 nm), where all samples absorb strongly ((~7000 M-’ cm-‘). [Rh(phpy),L]’ complexes have luminescence origins which lie about 2800 cm-’ higher in energy than the corresponding [ Rh (thpy ) *L] + complexes. This is illustrated in spectrum (a) for L=en. The luminescence of the [Rh(phpy)zL] + powders doped with 1% of [Rh(thpy)2L]+ (spectra (b), (c) and (d)) is a superposition of host and guest luminescence. In spite of the low concentration of the guest complexes, about 50% of the total luminescence originates from them in the spectra (b) and (c). This is attributed to energy migration through the host crystal to the guest traps [ 111. In the dimer spectrum (fig. Id) the situation is more extreme, with more than 95% of guest luminescence. Energy migration is obviously much more efficient in this purely molecular crystal. The linewidth of the guest luminescence may

A. Zilian et al. /Chemical Physics I74 (I 993) 513-523

a)

neat [Rh(thpy)ten]PFb

[Rh(thpy),-u+(thpy),Rhl in lRh(phpy),-CI-CI,-(phpy)2Rhl

d) _

1”“1”“1”“1”“1”’ 14000

515

16000

t I”“l”‘ll”“l’ll

16000

20000

22000

wavenumber (cm-‘) Fig. I. Luminescence spectra at 5 K of neat and mixed crystals containing [ Rh (thpy )2L] + and [ Rh( phpy),L] + complexes. All five samples were excited at 363.8 nm. The arrows indicate the positions of the laser used to excite the corresponding luminescence spectra of fig. 2. In trace (c) energy ladders are used to mark overtones and combination bands of the five most intense fundamental vibrational sidebands.

change considerably compared to the neat material. We attribute this to matrix effects leading to more or less inhomogeneous broadening. The best result was achieved for [Rh( thpy),bpy] + doped into [ Rh(phpy),bpy ]PF, where we already know from single crystal absorption spectroscopy and from Xray diffraction that the crystal has very little disorder [ 9, lo]. We can take advantage of this situation since the [Rh(thpy)*L]’ complexes substitute for [ Rh( phpy ),L] + complexes having the same charge and a similar shape. In fig. 2 selectively excited luminescence spectra at 5 K of the guest complexes [ Rh (thpy )*L] + in their respective [ Rh ( phpy ),L ] + matrices are compared with the Raman spectrum of neat [Rh(thpy)zbpy]PF,. The excitation energies are indicated as arrows in fig. 1. These excitation energies lead to considerable narrowing of the luminescence lines. The line pattern of the spectra in traces (a), (c) and (d) is due to vibrational sidebands which can be ascribed to ligand internal modes covering energies from 3071

cm- ’ (very weak C-H stretching mode not shown in fig. 2) down to less than 100 cm-‘. As in ref. [ 181 the intense Raman peak around 80 cm- ’ can be attributed to torsional motions of the ligands. Between 630 and 1604 cm-’ ring deformations, C-C and CN stretching vibrations mixed with a decreasing amount of C-C-C and an increasing amount of CC-H wagging contributions, are observed. Rhodium-ligand modes are expected to lie between 180 and 400 cm- ’ [ 191, where we find many low-intensity vibrational sidebands. Sidebands of higher intensity attributed to metal-ligand vibrations are indeed observed in the corresponding iridium complexes [ 201. As seen in figs. 1 and 2 all the prominent sidebands can be assigned to fundamentals or combinations of the five modes l-5 whose energies are listed in table 1. In fig. lc energy ladders are used to indicate the positions of overtones and combination bands. The vibrational energies of the spectra shown in fig. 2 are identical to within + 3 cm-‘, which corresponds to less than ?I 0.5% of the frequencies.

516

A. Zilian et al. /Chemical Physrcs I73 (1993) 513-523

In [ Rh(thpy),bpy] + doped [Rh(phpy)z(c) ), where the natural linewidth is observed,andin [Rh(thpy)zen]+ (trace (a)),where the excitation light is resonant with the luminescent origin, we attribute this to electron-phonon coupling. On the other hand, the narrowed spectrum of [Rh(thpy)2-u-C12-(thpy)2Rh] (trace (d)) is excited well above the luminescent origin and there are additional sources for the broad intensity [ 13 1. The traces (d) in figs. 1 and 2 show that the [Rh(thpy)zu-CIZ-(thpy)zRh] luminescence is broad under UV excitation and that it can be partially narrowed with the appropriate choice of the exciting light. This reflects the microscopic variation of the complex geometry in the macroscopic crystal. The high resolution of the spectra shown in fig. 2 enables us to investigate how much the vibrational structure is influenced by matrix effects or by the chemical composition of the complex. The vibrational structure in the luminescence spectra of [Rh(thpy)2bpy]+ doped either into [Rh(phpy)zbpy]PF6 (fig. 2c) or into PMMA [ 111 differs only slightly from the Raman spectrum of neat [Rh(thpy)zbpy]PF6 (fig. 2b and table 1). Thus the matrix has only a minor effect on the vibrational energies. In the present work we also varied the chemical composition of the [Rh(thpy),L]+ complexes. While the two thienylpyridine ligands always remained unchanged, the third ligand 2,2’-bipyridine was replaced by ethylenediamine or by the bridging chloride ions in the case of the binuclear complex. Looking at the highly resolved luminescence spectra in fig. 2 and table 1 we recognize that the third ligand sideband.

bpy ] PF6 (trace

a)

IRh(thpy),enl+

b) IRh(tbpy)ZbpylPr’,

RUIXU~

c) lRh(Wy),bpyl+ NJ iRb(pbpy)2bpyll’l’

b

4 IRh(thpy),-~-Cl,-(lhpy)zRhI iniRh(phpy),-~-Cl,-(phpy),RhI

II -2000

0 8 0 In’8

)I8

-1500

-1000 wavenumber

1’0

II -500

0 q IlO

L,

0

(cm?)

Fig 2. Selectively excited luminescence spectra (at 5 K) of mixed crystals (a, c. d), as indicated, and a Raman spectrum of neat [Rh(thpy)rbpy]PFs. Theexcitationenergiesare 18863 cm-’ (a), 15449 cm-’ (b) and 19430 cm-’ (c, d ). The luminescence spectra are drawn on a scale relative to their electronic origin.

The relative intensities of corresponding lines are comparable in all luminescence spectra (traces (a), (c) and (d) in fig. 2) and are distinct from the intensities in the Raman spectrum. In the spectra (a), (c) and (d) each line is accompanied by a broader

Table 1 Fundamental vibrational sideband energies (in cm- ’ ) of [ Rh (thpy ) abpy ] + and [ Rh (thpy ) 2-u-Clz-(thpy ),Rh ] luminescence spectra. The estimated accuracy is rt 2 cm-’ Assignment

lRh(thpy),hpyl+ ‘)

1 &ring) 2 &ring) 3 6( ring) 4 u(C-C, C-N) 5 v(C-C, C-N) u(C-H) v(C-H)

376 630 711 1399 1476 3022 3071

316 629 711 1393 1482

‘) [Rh(thpy),bpy]+ in [Rh(phpy)abpy]PFs at 5 K, see figs. lc and 2c. ‘) [Rh(thpy)+-Cl,-(thpy),Rh] in [Rh(phpy)2-u-Clz-(phpy)2Rh] at 5 K, see fig. 2d.

A. Zilian et al. /Chemical Physics I74 (I 993) 513-523

L has only a minor influence on the nature of the lowest excited state. All [ Rh (thpy ) zL ] + complexes have the same vibrational energies within + 3 cm-‘. 3.2. Excited state vibrational structure The high-resolution excitation spectra of [ Rh (thpy ),L] + complexes provide access to the vibrational energies of their lowest excited triplet state. Fig. 3 shows the first few hundred wavenumbers of the excitation spectra (at 5 K) of [ Rh (thpy ) ,bpy ] + in PMMA (trace (a)), in a [Rh(phpy)zbpy]PF6 crystal (trace (b)), and of [Rh(thpy)z-p-C&-

(tb-v)2Rhl doped in a [Rh(phpy)2-~-C12-(phpy)2Rh] crystal. Spectrum (b) was excited with a monochromatized tungsten lamp and detected through a cut-off filter below 17500 cm-‘. The natural linewidth is 5 cm-‘. For the spectra (a) and (c) a nar-

I

C

IRhWw),bpyl+

C)

I

IRh(lhpy),-~-Clz-(thpy)2Rhl

18800

19000

19200

f

19400

19600

wavenumber(cm-‘) Fig. 3. Origin region of the excitation spectra (at 5 K) of

l~(t@y)&pyl+

(a, b) and [Rh(thpy)2-~-Cl2-(thpy)2Rhl (c)

complexes doped into PMMA (a) or a host crystal (b, c). Traces (a) and (c) are narrowed excitation spectra and the luminescence was monitored in the line C (a: 19150 cm-‘, c: 18745 cm-’ ). Trace (b) shows the natural linewidth of the spectrum.

517

rowing technique [6-81 was used, i.e. the luminescence was monitored at the position C (the luminescence origin) while the pulsed dye laser was scanned. At the position C the excitation and detection frequencies coincide. The effects of scattered laser radiation were minimized with the help of a gated photon counter. This procedure leads to less than 20% uncertainty in the intensity of line C. Neither spectrum (a) nor (c) is corrected for the laser intensity, which drops markedly above 19500 cm-‘. A star in traces (b) and (c) at 5 14.5 nm marks the position of the Ar+ laser used to excite the corresponding luminescences (fig. 2, traces (c) and (d) ). In all of the three spectra we observe in addition to the origin C a second origin designated D which lies about 158 cm-’ higher in energy. For [Rh(thpy)zbpyl’ doped [Rh(phpy)zbpylPF6 origin C lies at 19208 cm-’ and origin D at 19364 cm-‘. The width of line D varies a great deal from one sample to another and is obviously not instrumentally limited. Traces (a) and (c) show spectra of samples with comparable inhomogeneous broadening. The origin D is one to two orders of magnitude broader than the origin C in these spectra. The line narrowing technique is not able to remove the broadening for both origins C and D. This means that the energy difference between the two excited states C and D varies within the inhomogeneous distribution, or, in other words, the two excited states C and D are not correlated. In [ Rh(thpy)2bpy] + doped [ Rh( phpy),bpy ] PF6 the origin D is well resolved and sharp, because the inhomogeneous broadening is much smaller in this crystal (fig. 3b). Fig. 4 shows a comparison of the vibrational sideband structure in the excitation spectra of the same samples as in fig. 3. The spectra (a) and (c) are not corrected for laser intensity. Whereas for trace (c) it was possible to work mainly in the flat regions of the dyes, the features around 600 cm-’ in spectrum (a) lie in a region of strongly changing laser intensity and were therefore recorded with two adjacent dyes. Spectrum (b) were excited with a halogen lamp in conjunction with a double monochromator and is corrected for its response. Fig. 4 shows that all the prominent sidebands can be assigned to nine fundamentals and their combination bands. An energy ladder constructed from the nine intense fundamentals (trace (b) ) is repro-

518

A. Zilian et al. /Chemical Physics 173 (1993) 513-523

b) [Rh(thpy)&py1 inNWw9,bpylPF,

1”

1000

IRh(phpy),-ll-cl,-(phpy),Rhl

1500

wavenumber (cm-‘) Fig. 4. Vibrational sidebands of the excitation spectra of fig. 3. The spectra arc aligned so that the origins C coincide at 0 wavenumbers. The energy ladders use the nine most intense vibrational sldebands of trace (b) and mark the corresponding fundamentals and combination bands.

duced in trace (a) (with the same intervals as in trace (b) ) to indicate the positions of overtones and combination bands. The vibrational energies of these fundamentals are listed in table 2 for the traces (b) and (c). Vibrational sidebands of the origin D are listedonlyfor [Rh(thpy)zbpy]+doped [Rh(phpy)r bpy ] PF6 (trace (b) ), where they are experimentally accessible. In the narrowed spectra of fig. 4 (traces (a) and (c) ) the origin D remains broad, which makes it impossible to accurately determine the corresponding vibrational energies. On the other hand the excitation spectra are thus simplified and the vibrational structure built on line C is more clearly visible. We note that in all the spectra of fig. 4 the vibrational sideband structure is due to the thienylpyridine ligands and extends to 2500 cm-’ above the origin designated C. The steep increase of absorption in trace (c) above 2500 cm-’ corresponds to the absorption edge of the [Rh(phpy)z-u-Clz-(phpy)lRh] host. We notice the near coincidence of corresponding vibrational energies in all the three samples. The only minor deviation is the pair of lines (around 600

cm - ’ ) instead of a single one in trace (c). As for the ground state we thus find that the vibrational energies of the lowest excited triplet state are largely independent of the matrix or the “spectator” ligand L in the complex (fig. 4 and table 2). The listed energies are identical to within k6 cm-’ (table 2), i.e. the variation is twice as big as for the ground state (section 3.1). In the case of [ Rh ( thpy)zbpy] + doped [ Rh ( phpy ),bpy ] PF6 (fig. 4b) the vibrational sideband energies of the origin D are up to 9 cm-’ smaller than those of state C (table 2). The values of both excited states for [ Rh (thpy ),bpy ] + lie well within the range of values obtained for the state C in the various complexes and matrices. As an illustration, table 2 shows that state D in [ Rh( thpy ),bpy ] + and state C in [ Rh (thpy ) 2-u-C12- ( thpy)zRh] have almost identical energies. As was shown in ref. [ lo], lines C and D can be ascribed to excited states with dominant 3x-x* character localised on two crystallographically inequivalent thpy- ligands. The vibrational sideband structure thus has contributions only from one ligand

A. Ziban et al. /Chemical Physics 174 (1993) 513-523 Table 2 Fundamental vibrational sideband energies (in cm-‘) of [ Rh(thpy)2bpy]+ estimated accuracy is k 2 cm-’ Assignment

and [ Rh(thpy),-p-Cl,-(thpy),Rh]

lRh(thpy)&nyl+ =)

519

excitation spectra. The

[Rh(thpy),-CI-Cl,-(thpy),Rhl d, origin C =)

16( ring) 2 6( ring) 3 G(ring) 4 u(C-C, 5 Y(C-C, 6 u(C-c, 7 u(C-C, 8 v(C-c, 9 u(C-C,

C-N) C-N) C-N) C-N) C-N) C-N)

origin C b,

origin D =)

371 581 689 1101 1207 1261 1288 1483 1544

380 575 685 1097 1264 1281 1476 1539

‘) [Rh(thpy)*bpy]+in [Rh(phpy)2bpy]PF,at 5K,seetig.4b. b, Vibrational sidebands of origin C. ‘) Vibrational sidebands of origin D. ‘) [Rh(thpy)2-p-(thpy)zRh] in [Rh(phpy)2-p-C1,-(phpy)zRh] ‘) Vibrational sidebands of origin C.

being excited, e.g. for transition C: [Rh(thpy)c(thpy),L]+-+ [Rh(thpy)E(thpy)pL]‘. Our absorption and luminescence spectra do not allow us to measure vibrational energies of the unexcited ligands. 3.3. Correlation ofground and excited state

vibrational structure

K

378 571 688 1091 1205 1257 1279 1472 1541

In fig. 5 the vibrational sideband structure of the 5 excitation and luminescence spectra of

[Rh(thpy)zbpyl’ doped [RhWwMwylPF~ is compared. Both spectra are plotted on an energy scale relative to the common origin C. Luminescence originates from C at all temperatures below 60 K, irrespective of the dopant concentration which was varied from 0.01 to 1 mole% The most prominent fundamental vibrational sidebands are numbered l5 in the luminescence spectrum and l-9 in the excitation spectrum, respectively. In both spectra we find a similar intensity distribution with the origin as the most intense line, which is typical for a symmetry allowed transition. In a first approximation a mirror relationship is expected for the two spectra. In the first 100 cm- ’ this is fulfilled both for the energies and intensities of the sidebands. At higher energies, a significant deviation from the mirror relationship is caused by the additional line D in the excitation spectrum. Except for this complica-

at 5 K, see fig. 2d.

tion, a pretty good mirror relationship is retained up to about 1000 cm-‘. In particular, the prominent sidebands 1, 2 and 3 show a reasonable correspondence of both energy and intensity in the two spectra, see fig. 5 and tables 1 and 2. In the energy range above 1000 cm-’ an unambiguous correlation of ground state and excited state vibrational frequencies is not possible. For example, there is a group of lines around 1200 cm-’ in the excitation spectrum which have no obvious counterparts in luminescence, see fig. 5. As shown in fig. 4 and table 2, the same lines are observed at almost the same energies relative to the origin line in the excitation spectrum of [ Rh(thpy)zp-CL(thpy),Rh] doped into [Rh(phpy)2-p-C12(phpy ),Rh], irrespective of the absolute position of the origin lines which differs by more than 450 cm-’ for the two systems (figs. 3b and 3~). Recently, the same group of lines could also be observed in the excitation spectrum of [ Ir( thpy),bpy ] + doped [ Rh( phpy ),bpy ] PF6. In the same spectrum the first )MLCT transition could be observed as a broad band approximately 2900 cm-’ above the narrow origin lines C and D of the ‘n-x* transition [ 201. We therefore conclude that the group of lines around 1200 cm-’ above the origin are fundamental vibrational sidebands and not electronic origins of higher excited ‘K-X* or a 3d-rr* states. The intensity distribution of vibrational sidebands

520

A. Zdian et al. /Chemical Phystcs 173 (I 993) 513-523

0

500

1000

1500

2000

2500

0

-500

-1000

-1500

-2000

-2500

wavenumber(cm~')

Fig. 5. Comparison of the vibrational sidebands of the excitation and luminescence spectra of [ Rh( thpy ),bpy ] + in [ Rh( phpy ),bpy ] PF6 (cf. figs. 2c and 4b). The spectra are aligned so that the common origin C lies at 0 wavenumbers. The energy ladders indicate the most intense fundamentals and their combination bands.

in an allowed transition is in a first approximation given by the Franck-Condon (FC) factors of the relevant totally symmetric modes [ 2 11. In the linear FC approximation ground and excited state have the same force constants and a mirror relationship of the vibrational sidebands between absorption and luminescence is expected. Using a single-configurational coordinate model, the intensity distribution in the totally symmetric vibrations can be described by the Huang-Rhys factor S,:

S,=n-I”, In-1

(1)

where i designates the mode and I,, the intensity of the nth line in the progression [ 221. In more complicated situations, for example when several progressions are convoluted or when mode coupling is present, the “time dependent” theory [ 231 offers computational advantages over the single-configurational coordinate model. The intensities I, and I,. of the prominent fundamental vibrational modes, relative to the origin C, have been determined for the luminescence and for

the excitation spectrum shown in fig. 5, respectively. They are in the range 0.1 l-O.38 and 0.07-0.37 for the ground state (I,) and the excited state (I,. ) vibrational modes, respectively. According to eq. ( 1) these are just the Huang-Rhys factors S,, and we note that they cover about the same range in excitation and in luminescence. Similar S, can be observed for n-x* transitions in various other systems [ 24-291. In the energy region below 1000 cm-‘, where a correlation between ground and excited state frequencies is possible for the [ Rh( thpy)abpy] + complexes, the corresponding intensities I, and 1,. are comparable in luminescence and excitation, see fig. 5. We therefore conclude that the linear FC term dominates the sideband intensity of these modes. A need for some modification arises from the fact that even in this energy range the modes 2 and 3 do not have equal frequencies in luminescence and excitation. While frequency v, remains unchanged, the frequencies v2 and v3 decrease by 7.8 and 3.1%, respectively, upon excitation. Hence there is no general decrease of frequencies in the excited state compared to the ground state, although an antibonding rc’ orbital is occupied. A sim-

521

A. Zilian et al. /Chemical PhysicsI74 (I 993)513-523 ilar behaviour was found for the lowest ‘x-z* transition of pentacene [ 241. In addition, our finding that a correlation of ground state and excited state vibrational frequencies is not possible above 1000 cm-’ for [Rh(thpy)zbpy]+ doped [ Rh (phpy ),bpy ] PF6 also shows that the linear FC approximation is too crude for this system. The deviations from a mirror relationship between the excitation and luminescence spectra of [ Rh (thpy ),bpy ] + can be understood by making use of a normal coordinate analysis which was recently done for both the ground state and the 3MLCT excited state of the Ru(bpy ):+ complex [ 30,311. A comparison of the 20 totally symmetric normal modes of a Ru-bipyridine moiety shows that there is no direct correspondence of modes in the ground and excited state. The potential energy distributions (PEDs) are comparable in the two states for totally symmetric vibrational modes with well separated energies in the region below 1000 cm- ‘. The density of totally symmetric modes increases above 1000 cm-‘, and between 1430 and 16 10 cm- ’ there are several modes with completely different compositions in the ground and excited state. The PEDs in this energy range are obviously very sensitive to a d-n* excitation. This means that a one to one correlation of modes in the ground and excited states is no longer possible, thus leading to deviations from a mirror relationship in the corresponding spectra. There exist 33 totally symmetric normal modes for the Rh-thienylpyridine moiety of the [Rh(thpy)2bpy] + complex. We believe that an analogous disparity in the mode composition is responsible for the differences of the ground state and excited state vibrational sideband pattern observed between 1000 and 1600 cm- ’ in fig. 5. In both X-X* (title complex) and d--A* excitations (Ru(bpy)$+ ) the lowest rc* antibonding orbital is populated. The resulting degree of reorganisation of electron density, which is reflected in the composition of the normal modes, should be comparable.

placement of the potential energy surface along totally symmetric normal coordinates, on the other hand, leading to another equilibrium geometry. Such bond length changes are reflected in the change of bond orders when promoting an electron from a rt to a 1z*orbital. For a d-x* excitation very similar bond length changes are expected in the ligand framework, since the bond orders undergo changes comparable to those for a x-rc* excitation. In order to work out the line nuances between the d-K* and the x--x* excitations we need a very precise description of the excited state geometry. Spectroscopic quantities such as the frequencies and intensities of vibrational sidebands in luminescence and absorption, respectively, contain valuable information about geometry changes upon electronic excitation. Various attempts have been made to use this information to obtain an estimate of the excited state geometry, see e.g. refs. [ 19,22,28,30-321. Various %--x” emitters such as bpy, phen and show highly resolved luminescence Rh(phen)z+ spectra [ 25-271. The intensities and frequencies for individual ground state vibrational modes can be easily determined. Yet the data set is not complete because the excited state vibrational structure was not accessible in these systems. The [ Rh(thpy)zL] + complexes studied in this work provide the advantage that the spin selection rule is sufficiently relaxed so that we can measure high-resolution luminescence and excitation spectra of the lowest triplet state without great difficulty. In the following we discuss the possibilities to derive an estimate of the excited state geometry from these spectra. Several empirical rules have been found which correlate frequency ratios v*/v with bond length changes [ 33 1. Badger’s rule for example relates the force constants of diatomic molecules to their equilibrium bond lengths r, [ 341. Herschbach and Laurie discovered a somewhat simpler relationship which gave a better lit to a larger set of data [ 35 ] :

3.4. Geometry of the lowest excitedstate

r,,

In the [ Rh(thpy),L] + complexes 3ir-?t* excitations promoting an electron from a bonding to an antibonding ligand orbital lie at lowest energy. A 3~-~* excitation results in a change of force constants and vibrational frequencies on the one hand and a dis-

Though this relation between differences and vibrational nally derived for diatomic equation has been used for atomic systems to make an

(A)-re,

(A)=-0.4081n(vZ/v1).

(2a)

equilibrium bond length frequencies was origimolecules, an analogous more complicated polyorder of magnitude esti-

522

A. Zilian et al. / Chemical Physics 173 (1993) 5 13-523

mate of the geometry change AQ, upon electronic citation along a normal coordinate: AQ, (A)=-0.408ln(v:/v,)

_

ex-

(2b)

However, the displacements along normal coordinates cannot be accurately translated into individual bond length changes in the absence of reliable normal-mode formulations. Furthermore, the application of eq. (2b) is limited to situations where the ground state and the excited state vibrational frequencies v and v*, respectively, correspond to normal modes with comparable mode compositions in both electronic states. The deviations from the mirror relationship show that this is not the case for the [Rh(thpy),bpy]+ system (see section 3.3) and therefore eq. (2b) cannot be used. However, an extension of eq. (2a) to different electronic states of a polyatomic system is possible if a normal coordinate analysis can be performed for both states. The elements of the PEDs are used to estimate “pseudodiatomic” stretching frequencies of a given bond and these “pseudodiatomic” stretching frequencies are then used to calculate individual bond length changes in a polyatomic system [ 311. This procedure has been applied to the charge transfer systems bpy/bpy-’ and Ru(bpy):+/(Ru(bpy)$+)* [ 19,3 11. It was concluded that the bond connecting the pyridine rings is stronger and thus shorter in the excited state and therefore responsible for the frequency increase of some modes. At the same time some of the bonds in the ring get weaker and thus longer. This description of the excited state geometry is consistent with bond order calculations on bpy [ 361 and on biphenyl [ 371. In order to determine these individual bond length changes for the Ru(bpy):+ / (Ru (bpy )z+ )* system, a normal coordinate analysis was done of both ground and excited states with a careful refinement of the forcefields. Twelve isotopomers of the Ru (bpy ):+ complex were used to derive the 35 parameters of the forcefields [ 30,3 1]. We have tried to perform a normal coordinate analysis for the [ Rh ( thpy)zL] + complexes using the vibrational frequencies of their fully protonated isotopomers. Since the lowest-energy excited state can be ascribed to the thienylpyridine ligands, we attempted to set up the force fields for the Rh-thienylpyridine moiety in analogy to the Ru (bpy ) :’ / (Ru (bpy ) :’ )* system, and to use the results of the normal coordi-

nate analysis of this system to derive initial values of the force constants for the Rh-thienylpyridine moiety. We hoped that with sufficiently accurate initial values it would be possible to refine the force constants even with the limited amount of data available for the ( fully protonated) [ Rh( thpy ),L] + complexes. However, we had to conclude that our data base was not sufficient. Many different isotopically labeled [Rh(thpy),L] + complexes would be needed to derive reliable force constants and accurate values for individual bond length changes. Another access to the excited state geometry is given by the Huang-Rhys factors in the Franck-Condon approximation. The displacement along a normal coordinate Q, is given by [ 22 ] I4l=J2s,

>

(3a)

[AQ,l (A)=5.804(d’ Jz’

(3b)

where p is the reduced mass (amu) and v, the vibrational frequency (cm-’ ). Eq. (3) has the same disadvantage as eq. (2b): the displacements along normal coordinates can only be translated into individual bond length changes if reliable PEDs of the different normal modes are known. Since a normal coordinate analysis could not be done, the PEDs are not known for the title complex. Therefore, we can give only an order of magnitude estimate of the displacements along normal coordinates for which comparable mode compositions are expected in the ground and excited state (i.e. for the frequency range below 1000 cm- ’ ) . Using eq. (2b) and the frequency ratios from tables 1 and 2 on the one hand, and eq. (3 ) and the intensity ratios from fig. 5 on the other hand, we obtain from both approaches distortions of the order of a few hundredths of an A. The distortions are of the same magnitude as those obtained for charge transfer transitions to a bpy ligand [ 19,31,32] and we find that the vibrational structure of the ligands is affected in a similar way.

4. Conclusions Often the term excited state geometry is used when dimensionless distortions from the ground state geometry, d (eq. (3a) ), are determined. Using the

A. Zilian et al. /Chemical Physics 174 (I 993) 513-523

Huang-Rhys factors S (eq. ( 1) ) this is straightforward in our case by virtue of the highly resolved spectra. In the case of resonance Raman or less resolved absorption and luminescence spectra, this requires more work, and the A parameter can be obtained only after intense fitting. Proceeding from this point, a determination of the excited state geometry measured in internal coordinates apparently is an even bigger undertaking. The crucial step is performing a reliable normal coordinate analysis. This problem, although treatable for small molecules, is currently untreatable for most others. In the best cases, the solutions are not unique and need further backing from molecular orbital calculations [ 381 or from a comparison with structural data of model compounds mimicking the excited state of the compound of interest [ 391. Other results [ 32,401 stand on much shakier ground. We sadly conclude that despite the excellent quality and the high information content of the spectra reported here we are not able to derive accurate information about changes of the molecular geometry in the excited electronic state. Acknowledgment We thank Mark Riley for his help with the normal coordinate analysis and Mirco Colombo for fruitful discussions. Financial support by the Swiss National Science Foundation is gratefully acknowledged. References [ 1] E. Krausz and J. Ferguson, Progr. Inorg. Chem. 37 (1989) 293. [2] H. Yersin, D. Braun, G. Hensler and E. Gallhuber, in: Vibronic processes in inorganic chemistry, ed. C.D. Flint (Kluwer, Dordrecht, 1989). [ 31 R. Schwarz, G. Gliemann, L. Chassot, P. Jolliet and A. von Zelewsky, Helv. Chim. Acta 72 (1989) 224. [4] M. Kirch, J.M. Lehn and J.P. Sauvage, Helv. Chim. Acta 62 (1979) 1345. IS] M.K. Nazeeruddm, P. Liska, J. Moser. N. Vlachopoulos and M. Graetzel, Helv. Chim. Acta 73 (1990) 1788. [ 6) A. Zilian. M.G. Colombo and H.U. Gtldel, J. Luminescence 45 (1990) 111. 171 M.G. Colombo, A. Zilian and H.U. Gtldel. J. Luminescence 48&49 ( 199 I ) 549. [ 81 M.G. Colombo, A. Zilian and H.U. Gtldel, J. Am. Chem. Sot. 112 (1990)4581. [9] A. Zilian, U. Maeder, A. von Zelewsky and H.U. Gtldel, J. Am. Chem. Sot. 111 (1989) 3855.

523

[lo] G. Frei, A. Zihan, A. Raselh, H.U. Gildel and H.-B. Bilrgi, Inorg. Chem., 31 (1992) 4766. [ I1 ] A. Zilian and H.U. Gildel, Inorg. Chem. 31 (1992) 830. [ 121 H. Riesen, E. Krausz, A. Zilian and H.U. Gildel, Chem. Phys. Letters 182 ( 199 1) 27 1. [ 131 A. Zilian and H.U. Gtldel, J. Luminescence 5 1 ( 1992) 237. [ 14 ] C.P.M. Giesbergen, R. Sitters, G. Frei, A. Zilian, H.U. Gtldel and M. Glasbeek, Chem. Phys. Letters 197 ( 1992) 45 I. [ 15 ] U. Maeder, T. Jenny and A. von Zelewsky, Helv. Chim. Acta 69 (1986) 1085. [ 161 U. Maeder, Inauguraldissertation Universitlt Fribourg, Switzerland ( I987 ) [ 171 E. Krausz, C. Tomkins and H. Adler, J. Phys. E 15 (1982) 1167. [ 181 M. Muniz-Miranda, E. Castellucci, N. Neto and G. Sbrana, Spectrochim. Acta 39A (1983) 107. [ 191 G.D. Danzer, J.A. Golus, D.P. Strommen and J.R. Kincaid, J. Raman Spectry. 2 I ( 1990) 3. [20] M.G. Colombo and H.U. Giidel, Inorg. Chem., in press. [ 2 1 ] I.S. Osad’ko, Soviet Phys. Uspekhi 22 ( 1979) 3 I 1. [22] R.B. Wilson and E.I. Solomon, J. Am. Chem. Sot. 102 (1980) 4085. [23] E.J. Heller, Accounts Chem. Res. 14 (1981) 368. [24] R.J. Carlson and J.C. Wright, J. Chem. Phys. 92 (1990) 5186. [25] N. Okabe, T. Ikeyama and T. Azumi, Chem. Phys. Letters 165 (1990) 24. [ 26 1K. Vinodgopal and W.R. Lcenstra, J. Phys. Chem. 89 ( 1985) 3824. [ 27 ] Y. Komada, S. Yamauchi and N. Hirota, J. Phys. Chem. 90 (1986) 6425. [28] D.M. Burland and G.W. Robmson, J. Chem. Phys. 51 (1969) 4548. [29] T. Huang, K.E. Rieckhoff and E.M. Voigt, Chem. Phys. 36 (1979) 423. [ 30) D.P. Strommen, P.K. Mallick, G.D. Danzer, R.S. Lumpkin and J.R. Kincaid, J. Phys. Chem. 94 (1990) 1357. [ 311 P.K. Mallick, D.P. Strommen and J.R. Kmcaid, J. Am. Chem. Sot. 112 ( 1990) 1686. [ 32 ] J.V. Caspar, T.D. Westmoreland, G.H. Allen, P.G. Bradley, T.J. Meyer and W.H. Woodruff, J. Am. Chem. Sot. 106 (1984) 3492. [33] H.B. Bhrgi and J.D. Dunitz, J. Am. Chem. Sot. 109 (1987) 2924. [ 341 R.M. Badger, J. Chem. Phys. 2 ( 1934) 128. [35] D.R. Herschbach and V.W. Laurie, J. Chem. Phys. 35 (1961) 458. [ 361 T. Schonherr, J. Degen, E. Gallhuber, G. Hensler and H. Yersin, Chem. Phys. Letters I58 ( 1989) 5 19. [37] S. Ramasesha, I.D.L. Albert and B. Sinha, Mol. Phys. 72 (1991) 537. [ 381 T. Kato, N. Muraki and T. Shida, Chem. Phys. Letters 164 (1989) 388. [39] M.H. Chisholm, J.C. Huffman, I.P. Rothwell, P.G. Bradley, N. Kress and W.H. Woodruff, J. Am. Chem. Sot. 103 ( 198 1) 4945. [ 401 P. Stein, M.K. Dickson and D.M. Roundhill, J. Am. Chem. Sot. 105 (1983) 3489.