Fluorescence excitation and excited state intramolecular relaxation dynamics of jet-cooled methyl-2-hydroxy-3-naphthoate

Chemical Physics 425 (2013) 177–184

Contents lists available at ScienceDirect

Chemical Physics journal homepage: www.elsevier.com/locate/chemphys

Fluorescence excitation and excited state intramolecular relaxation dynamics of jet-cooled methyl-2-hydroxy-3-naphthoate Annemarie McCarthy, Albert A. Ruth ⇑ Physics Department & Environmental Research Institute, University College Cork, Cork, Ireland

a r t i c l e

i n f o

Article history: Received 25 July 2013 In final form 30 August 2013 Available online 7 September 2013 Keywords: Methyl-2-hydroxy-3-naphthoate Fluorescence excitation Supersonic jet Vibronic structure Intramolecular proton transfer Conformer

a b s t r a c t Two distinct S0 ? S1 fluorescence excitation spectra of methyl-2-hydroxy-3-napthoate (MHN23) have been obtained by monitoring fluorescence separately in the short (410 nm) and long (650 nm) wavelength emission bands. The short wavelength fluorescence is assigned to two MHN23 conformers which do not undergo excited state intramolecular proton transfer (ESIPT). Analysis of the ‘long wavelength’ fluorescence excitation spectrum, which arises from the proton transfer tautomer of MHN23 indicates an average lifetime of s P 18 ± 2 fs for the initially excited states. Invoking the results of Catalan et al. [J. Phys. Chem. A, 1999, 103, 10921], who determined the N tautomer to decay predominantly via a fast non-radiative process, the limit of the rate of intramolecular excited proton transfer in MHN23 is calculated as, kpt 6 1  1012 s1. Ó 2013 Elsevier B.V. All rights reserved.

1. Introduction The present work on the fluorescence excitation spectra of jetcooled methyl-2-hydroxy-3-naphthoate (MHN23) supplements earlier work on the excited state intramolecular proton transfer of the jet-cooled napthol derivatives 1-hydroxy-2-naphthaldeyde [1] and 2-hydroxy-1-naphthaldehyde [2]. MHN23 is the naphthalene equivalent of methyl salicylate (MS), a molecule which exhibits one of the most extensively studied excited state intramolecular proton transfer processes [3–6]. MHN23 undergoes an excited state intramolecular proton transfer (ESIPT) analogous to that which occurs in MS. However, MHN23 has not been as extensively studied as MS and its photophysics is still not fully understood. Bergmann et al. [7] conducted the first study of MHN23 in solution and observed the S1 absorption band at 375 nm in dioxane solution. The authors established the presence of an intramolecular hydrogen bond in MHN23. Dual fluorescence emission in MHN23, was first observed by Naboikin et al. [8]. In Ref. [8], an emission at 410 nm and a Stokes shifted (9000 cm1) emission at 650 nm were observed. The long wavelength emission was subsequently ascribed to the keto tautomer of MHN23 (Fig. 1) by Woolfe and Thistlethwaite [9]. Law and Shoham [10] reported an intrinsic fluorescence quantum yield of 0.7% of MHN23 in methylcyclohexane at room temperature. Catalan et al. [11] have conducted the most recent study of MHN23 and its structural isomers methyl 1-hydroxy2-naphthoate and methyl 2-hydroxy-1-naphthoate. Absorption, ⇑ Corresponding author. Fax: +353 21 4276949. E-mail address: [email protected] (A.A. Ruth). 0301-0104/$ - see front matter Ó 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.chemphys.2013.08.017

emission, and excitation spectra of MHN23 in solution and in the gas phase were measured. A theoretical study of the molecule was also conducted. The DFT B3LYP/6-31G** ground state energies of four MHN23 conformers were computed and the N tautomer (see Fig. 1) was found to be the dominant ground state species. Upon excitation to the S1 state, ESIPT in the N tautomer is driven by increases in the acidity and basicity of the hydroxyl and carbonyl groups, respectively. Catalan et al. [11] concluded that proton transfer occurs with a yield of only ca. Upt  0.018, across a potential energy barrier of 1.5 kcal/mol in the S1 state. Radiative decay of the keto tautomer produces the observed 650 nm fluorescence band. A biexponential decay with lifetimes of 14.5 ns and 2.9 ns were observed by Catalan et al. [11] in the short wavelength fluorescence emission at 410 nm. The two fluorescence components were tentatively assigned to the HB and NHB conformers (see Fig. 1). No component of the short wavelength fluorescence was ascribed to the N tautomer which was explained to decay predominantly via an efficient non-radiative process. Intersystem crossing to a triplet state was proposed as a possible N tautomer decay mechanism. MHN23 has not previously been studied in a supersonic jet. In this publication, the photophysics of MHN23 is investigated, through fluorescence excitation measurements of jet-cooled MHN23. Two distinct excitation spectra were obtained by monitoring the fluorescence in the short (410 nm) and long (650 nm) wavelength emission bands. A full description of the experimental apparatus was presented in [1] and will not be repeated here. Experimental details pertaining to specific measurements are also given in the figure captions.

178

A. McCarthy, A.A. Ruth / Chemical Physics 425 (2013) 177–184

wavelength [nm] 380

378

376

374

372

30

O(HB) 25

relative intensity

O(NHB-1) 20

15

O(NHB-2)? 10

5

0 26200

26400

26600

26800 -1

wavenumber [cm ] Fig. 3. Low energy part of the S1 S0 fluorescence excitation spectrum of MHN23 (short wavelength fluorescence, 410 nm). The positions of the origins of the HB (26743.7 cm1) and NHB-1 (26372.7 cm1) are marked with arrows. A weak feature at (26356.0 cm1) at a slightly lower energy than O(NHB-1) may be the origin of the NHB-2 conformer (tentative). Solid line between 26200 and 26450 cm1: Lorentzian fit to data (see text).

Fig. 1. Chemical structures of the MHN23 conformers, the nomenclature was adapted from Ref. [11] and also the relative separation of calculated ground state energies (DFT B3LYP/6-31G**) of the conformers, which will provide arguments in the spectral assignments and discussion in Section 2. The N tautomer is the dominant ground state species and the proton transfer (PT) tautomer (keto) is formed after ESIPT of the N tautomer [11]. (N)HB = (non)hydrogen bonded. The decay rates kr (radiative), knr (non-radiative) and kpt (proton transfer) of different processes which are discussed in Section 2.3 are also shown.

wavelength [nm] 380

375

370

365

360

120

410 nm

relative intensity

100

80

60

40

20

0 26250

26500

26750

27000

27250

27500

27750

-1

wavenumber [cm ] Fig. 2. S1 S0 excitation spectrum of the S1 ? S0 short wavelength fluorescence (410 nm) of jet-cooled MHN23. The fluorescence was collected through a 435 nm (FWHM 40 nm) interference filter. The dashed line indicates the relative excitation power (laser power 1.5 mJ per pulse in maximum at 27250 cm1 (dye: frequency doubled emission of Styryl 8). The spectrum has been scaled to the varying laser power [1]. Carrier gas was argon at a stagnation pressure of 760 mbar.

MHN23 (C12H10O3, m.p. = 73–75 °C, minimum purity 98%) was purchased from Sigma Aldrich and used without further purification. 2. Fluorescence excitation spectra of MHN23 2.1. Short wavelength emission The excitation spectrum of the S1 S0 electronic transition of jet-cooled MHN23, obtained by monitoring the short wavelength fluorescence (410 nm, collected through an interference filter centered at 435 nm) is shown in Fig. 2. The excitation spectrum was measured from 24096 to 27778 cm1 with spectral resolution of 0.07 cm1 (FWHM of laser bandwidth) [1]. Spectral features were observed between 26300 cm1 and 27778 cm1. The spectrum consists of two distinct patterns of narrow lines with different intensities and a broad (‘‘humped’’) structure, all following similar progressions as discussed in the next section (cf. also Fig. 3). The observed narrow features were assigned to the HB and one of the NHB ground state conformers of MHN23. The nature of the broad structure is not clear as of yet. In the following, the vibrational assignments will be described and discussed on basis of our own calculations and the results by Catalan and co-workers [11]. 2.1.1. Vibrational assignments The strongest spectral feature occurs at 26744 cm1 and is likely to be an origin of one of the MHN23 conformers (the relative intensity of the spectrum was normalized to 100 arb. units based on this line, Fig. 2). A series of weaker features appear on the low energy side of this strong origin (see Fig. 3), which did not change when different carrier gases (argon or xenon rather than helium) were used for the supersonic expansion. Therefore these weak features cannot be based on van der Waals complexes of MHN23 with atoms of the expansion gas. The expected small changes in rotational temperature caused by different carrier gases at similar stagnation pressures did also not change the shape of the weak

179

A. McCarthy, A.A. Ruth / Chemical Physics 425 (2013) 177–184

Table 1 S0 excitation spectrum of MHN23 (monitoring the short wavelength fluorescence, 410 nm). The excess energy (in wavenumbers) with Vibrational assignment of the S1 respect to the HB and NHB-1 origins are listed in columns 3 and 4, respectively. Features assigned to NHB-1 are in bold face as opposed to HB features. k (nm)

Relative intensity

HB

NHB-1

mexp (cm1)

mexp (cm1)

massign (cm1)

mexp  massign (cm1)

46 148 134 176 215a 258 285a 306 322a 310

8 24 27 5 2a 16 4a 7 0a 44

m014 m06 þ m010

433 474a

7 1a

2m001

96 146 148 192 568 616 626a 264 276a 290a

35 1 8 20 1 2 0a 12 0a 2a 4 5a 5

Assignment

379.42 379.18 378.40 377.40 376.89 376.60 376.14 375.29 375.18 374.74 374.61 374.16 373.92 372.96 372.50 372.16 372.10 371.91 371.75 371.53 371.20 370.50 370.39 370.10

6 14 8 6 15 7 6 8 8 15 11 8 100 7 11 10 20 91 28 24 6 10 10 23

388 371 317 247 211 190 158 98 90 58 50 18 0 69 102 126 131 145 156 172 196 247 255 276

17 0 54 124 161 181 213 274 281 313 322 354 (371) 440 473 498 (502) (516) (528) (543) 567 618 626 (647)

NHB-2 origin? NHB-1 origin 2m001 2m002

369.93

52

288

(660)

369.82

21

296

668

2m005 or m06 þ 2m002

369.60 369.46 369.14 369.01 368.94 368.72 368.56 368.16 367.97 367.88 367.83 367.65 367.49 367.27 367.05 366.25 366.18 366.11 366.02 365.96 365.91 365.79 365.70 365.58 365.41 365.38 365.25 364.94 364.80 364.48 364.36 364.28 364.19 364.03 364.00 363.83 363.77 363.75 363.67 363.48 363.36

53 11 7 17 9 6 9 34 48 40 15 60 28 13 15 26 20 68 26 31 38 13 53 21 32 34 11 9 12 11 14 16 44 14 28 14 51 53 16 32 10

313 323 346 356 361 377 389 418 432 439 442 456 468 484 500 560 565 570 577 581 585 594 601 610 623 625 634 658 668 692 701 707 714 726 728 741 746 747 753 768 777

(684) 694 718 (727) 732 748 760 (790) (804) (810) (814) (827) (839) (855) (872) 931 (936) (942) (948) (953) (957) 966 (972) 981 (994) (996 (1006) 1029 (1040) 1064 1073 (1079) (1086) 1098 1100 1113 1117 1119 (1125) (1139) (1149)

m010

292 301a 308

m011

340

16

420a 434a 437 443a 457a 469a 485a 500a

2a 2a 2 1a 1a 1a 1a 0a

m018 4m06 m06 þ m014 m08 þ m010

564a 570 578a 583a 588a

1a 0 1a 2a 3a

m010 þ 2m06

602a

1a

m020 2m010 m08 þ m011

614 625a 631a

8 0a 3a

m10 + m11

668a

0a

3m06 þ m08 m018 þ m06

710a 715a

2a 1a

2m010 þ 2m001 2m010 þ m06 2m010 þ 2m002

756a 770a 781a

2a 2a 4a

m06 m002 þ m004 2m001 þ m06 m08 2m002 þ m06 m010 2m06 m011 HB Origin

m06 2m002 m002 þ m003 m018 m020 2m010 m08 or 2m001 þ m06 2m06

m06 þ m08 3m06 m014 m010 þ 2m001 m010 þ m06 m010 þ 2m002 m010 þ ðm002 þ m003 Þ m06 þ m011 2m06 þ m08

(continued on next page)

180

A. McCarthy, A.A. Ruth / Chemical Physics 425 (2013) 177–184

Table 1 (continued) k (nm) 362.93 362.28 361.98 361.86 361.74 361.59 361.27 360.66 360.42 360.33

Relative intensity 11 26 26 36 13 11 7 8 9 11

HB

NHB-1

mexp (cm1)

mexp (cm1)

809 859 882 891 900 911 936 983 1001 1008

1181 1230 (1253) 1262 (1272) 1283 (1307) 1354 1373 1380

Assignment

massign (cm1)

mexp  massign (cm1)

2m014

878a

4a

2m010 þ m08

901a

1a

3m010

938a

2a

a Assignment based on a combination of lower energy features. massign is the sum of the experimental wavenumbers of the component vibrations. Otherwise, massign is the calculated vibrational wavenumber in S0 shown in Table S1 of the Supplementary material (DFT B3LYP/6-31G).

features. As the weak features’ intensities did not vary significantly relative to the strongest line (at 26744 cm1) upon changing the carrier gas backing pressure, hot bands were also discounted as the source of these features. It was noted that the vibronic structure based on the line at 26373 cm1 was very similar to that based on 26744 cm1. Since the computed vibrational modes of the MHN23 conformers are comparable (see Supplementary material, Table S1) the excitation spectra of the conformers can be expected to share some common features. Moreover, in Catalan et al. [11] the short wavelength emission was found to decay biexponentially with the associated times of 14.5 and 2.9 ns. The former was assigned to the HB rotamer whereas the latter was potentially (and tentatively) associated with one of the NHB forms. In ab initio calculations the oscillator strength of the p–p* transitions of all conformers were found to be of comparable strength (Table 4 in Ref. [11]). The intensity ratio of the two components in the biexponential fluorescence decay, Iweak/Istrong = 10/90 = 0.111 [11], agrees approximately with the intensity ratio of the ‘weak’ and ‘strong’ origins, I26373/I26744 = 0.138 in the excitation spectrum (Fig. 2). Hence the line at 26744 cm1 was assigned to the S1 S0 origin of the HB rotamer, whereas the one at 26373 cm1 was attributed to that of one of the NHB forms (notably the NHB-1 conformer; cf. next section and Table 1). The features of the ‘short wavelength’ spectrum are listed in Table 1. For the HB conformer a significant number of lines were based on the same vibrations. The assignments were done with the aid of a DFT B3LYP/6-31G** vibrational frequency calculation (see Supplementary material). The totally symmetric vibrations m06 and m010 both feature prominently in the spectrum and the harmonic progressions m06 ; 2m06 ; 3m06 ; 4m06 and m010 ; 2m010 and 3m010 were established. In addition, a number of combinations bands appear to be built on m06 ; 2m06 ; m010 and 2m010 . The totally symmetric (A0 ) vibrations m08 ; m011 ; m014 ; m018 and m020 , and three even number combinations of non-totally symmetric (A00 ) modes ð2m001 ; 2m002 and m001 þ m003 Þ have also been assigned. In Fig. 3 the low energy part of the spectrum containing the weaker features is shown. A very weak feature (rel. intensity 6 arb.u.) appears at 26356 cm1 (17 cm1 below the NHB-1 origin) and might be tentatively assigned as the S1 S0 origin of the other NHB conformer (see Section 2.1.2). A small number of very weak features (e.g., at 26479 cm1; corresponding to a vibrational energy of hc  123 cm1) could also be based on this feature. An underlying ‘‘humped’’ structure is also evident in Fig. 3. The broad structure appears to originate at 26351 cm1 and follows a progression that is based on intervals of 150–160 cm1. The nature of the corresponding broad features is still unexplained and several possibilities are considered here: (i) The interval of hc  155 cm1 corresponds approximately to the energy of the m06 vibration in the NHB-1 (161 cm1) and HB (145 cm1) conformers. Therefore at higher energies

the underlying structure could be simply attributed to the coincidence and overlap of a significant number of spectral features (from both HB and NHB conformers) in the same energy regions. However, the NHB conformer origin line also appears to exhibit an underlying broad ‘hump’. This broad feature cannot be accounted for by the co-incidence of HB and NHB spectral features, as both origins lie at higher energy. (ii) The ‘humped’ structure resembles the typical appearance of spectra that are governed by strong vibronic coupling between electronic states. In the classic case of naphthalene, the S2 origin is not well defined due to the strong vibronic coupling between S1 and S2, producing a complex structure in the S2 region [12]. As no spectral features are observed at wavenumbers smaller than ca. 26350 cm1, a lower lying state to which the observed state couples, would necessarily be ‘dark’ in nature [13]. The vibronic coupling mechanism would be expected to split the observed origins and vibronic bands into a number of features according to the density of dark states at the respective energies. The weak feature at 26356 cm1, which was tentatively assigned as the other NHB origin, could then instead be resulting from the vibronic coupling mechanism. However, the existence of a low lying dark state is not well supported, except for the fact that radiationless decay is expected to be the dominating relaxation pathway of the N tautomer according to Refs. [10,11]. (iii) If the progression of the broad features is not due to overlapping HB and/or NHB lines (see consideration (i)), one might consider the N tautomer to be the carrier of these bands. The N tautomer can be expected to have a vibrational structure similar to that of the HB and NHB conformers and thus a similar progression may be based on the N tautomer even though a well resolved vibronic structure cannot be expected owing to the expectedly short S1 lifetime of the N tautomer. A prompt fluorescence from the N tautomer was not observed by Catalan et al. [11], because the relaxation of the N tautomer was found to be governed by fast nonradiative mechanisms, among which ESIPT to the PT tautomer only constituted a minor relaxation pathway (yield of maximally 1.8% – see Section 2.3). The occurrence of a dual fluorescence (from the N and PT tautomer) was however not entirely ruled out in Ref. [11] since the existence of a double well potential along the O–H stretching coordinate with low energy barrier (1.5 kcal/mol) was deemed likely. The broadness of the feature in our excitation spectrum would in any case be indicative of a very short-lived species. The expectedly low fluorescence yield of the N tautomer could be (at least partially) compensated by the higher ground state population in comparison to the other conformers. According to Table 1 in Ref. [11] the most stable form of MNH23 is the N tautomer whose concentration is 66 times higher than that

A. McCarthy, A.A. Ruth / Chemical Physics 425 (2013) 177–184

wavelength [nm] 390

1.0

385

380

375

370

365

360

650 nm

relative intensity

0.8

0.6

0.4

0.2

0.0 26000

26500

27000

27500

-1

wavenumber [cm ] Fig. 4. Thin solid line: S1 S0 excitation spectrum of the S1 ? S0 long wavelength fluorescence of jet-cooled MHN23. The fluorescence was collected through a 650 nm (FWHM 80 nm) interference filter. Thick solid line: Smoothed spectrum by appropriate Fourier transform filter. Dashed line: relative excitation power (conditions as described in caption of Fig. 2); the spectrum has been scaled to the varying laser power [1].

of the HB rotamer. The area under the first broad feature at 26351 cm1 is only ca. 1.3 times smaller than the area under the HB origin at 26744 cm1, which would imply either a higher quantum yield of the N tautomer than expected or a larger population, or both. Moreover, in a supersonic jet (as opposed to solutions or static gas mixtures) the non-radiative decay of excited species is exclusively based on intramolecular relaxation, which may result in a notably larger ratio of the radiative and non-radiative rate coefficients. Based on these considerations the broad structure in the spectrum could potentially be assigned to the N tautomer. However, there are two strong arguments against this interpretation of the broad ‘humped’ structure in the spectrum: (a) The spectral position of the onset of the long wavelength fluorescence at ca. 25700 cm1 (see Fig. 4 in Section 2.2) is approximately 650 cm1 below the first hump in the short wavelength excitation spectrum. Assuming the long wavelength fluorescence to be due to ESIPT of the N tautomer, the onset of the short wavelength excitation spectrum would be expected to be approximately the same, if the N species contributed to it. Unless only vibronically excited N states are observed in the short wavelength excitation spectrum, for which non-radiative decay is particularly prohibited owing to the nature of the excited vibration – this scenario appears unlikely since the proton transfer rate would have to be dependent on the excess energy above the S1 origin. The shift between the onsets of the short- (410 nm) and longwavelength (650 nm) excitation spectra was also observed in solution [10,11] (ca. 1200 cm1) and in a static gas [11] (where it was significantly smaller than 1200 cm1), corroborating that different rotameric species of MHN23 contribute to the spectra. (b) The N tautomer would have to have a radiative decay rate constant of the order of ca. 1  1010 s1 in order to become observable in the spectrum, if the findings of Ref. [11] are correct. This is by a factor of 102–103 too large to be believable.

181

2.1.2. Stability considerations Considerations concerning the relative stability and electronic energies of the conformers of MHN23 are important in the context of the above assignments because only species whose populations in the jet are sufficient for the fluorescence to be observable (for a given yield) are contributing to the short wavelength excitation spectrum. We assume that conformer relaxation upon expansion in the jet is ineffective. This assumption is plausible as the energy barrier between the stability regimes of the conformers is large enough [11] so that the jet is essentially freezing the high-temperature mole-fractions of the conformers [14]. If conformer changes take place upon expansion the population of more stable species may be increased. According to Ref. [11] the most stable form of MNH23 is the N tautomer followed by the HB conformer and the two NHB conformers; the S0–S1 transition strengths of all conformers are comparable as calculated by Catalan et al. [11]. Unfortunately, ab initio calculations of conformational energies are inherently difficult and carry substantial errors. As argued in Ref. [2] a reasonable accuracy of calculations at the DFT B3LYP/631G** level is approximately ±30%. We calculated the free energy separation of the HB and NHB-1 conformers (at 343 K, 1 atm) to be (6 ± 2) kcal/mol at this level of theory, which represents a relative population ratio of 7000:1. The 30% uncertainty implies population ratios between ca. 370:1 and 135000:1. Assuming the absorption cross section and/or the fluorescence quantum yield of the NHB-1 species to exceed that of the HB tautomer, it is reasonable to assign the spectral features at 26373 cm1 to the NHB-1 rotamer. It is interesting to note that Catalan et al. [11] also studied two fluorescing conformers of methyl 1-hydroxy-2-naphthoate (MHN12), a structural isomer of MHN23. The relative fluorescence intensities of the two emitting species was 0.176 (cp. with the ratio stated in the previous section) and they were assigned as the HB and N tautomers of MHN12. The computed B3LYP/6-31G** free energy separation, at 298.15 K and 1 atm, of the HB and N tautomers of MHN12 was 4.1 kcal/mol, which yields a population ratio of [N]:[HB] = 1000:1 (at 298.15 K) [11]. Experimentally the population ratio of the conformers was found to be [N]:[HB] = 96:1. This ratio is of the same order of magnitude as the minimum population ratio computed for the HB and NHB-1 conformers of MHN23, illustrating the feasibility of observing the NHB-1 conformer of MHN23. As the energy separation of the rotamers NHB-1 (E = 11.5 kcal/ mol) and NHB-2 (E = 10.7 kcal/mol) is very small (i.e., DE = 0.8 kcal/mol  280 cm1), the NHB-2 fluorescence may also be observed in the excitation spectrum, supporting the rather tentative assignment of the feature at 26356 cm1 to NHB-2. Variation of the experimental sample temperature should allow for the adjustment of the relative populations of the HB and NHB-1 conformers and the improvement of signal to noise ratio in the region of the NHB-1 origin. Better resolution in the excitation spectrum in this region and/or measurement of single vibronic level fluorescence may yield insight into the nature of the observed structure of the NHB-1 origin line. 2.2. Long wavelength emission The excitation spectrum of the S1 S0 electronic transition of jet-cooled MHN23, obtained by monitoring the long wavelength fluorescence (650 nm) is shown in Fig. 5. The prompt fluorescence was collected through a 650 nm interference filter (FWHM 80 nm). The excitation spectrum was measured from 25630 to 27760 cm1 (taken with 2500 samples per data point). As the excitation spectrum was measured under the same experimental conditions as the short wavelength fluorescence excitation spectrum (Fig. 2), the broad nature of the spectrum is not a result of

182

A. McCarthy, A.A. Ruth / Chemical Physics 425 (2013) 177–184 Table 2 Results of the vibrational line fitting procedure for the long wavelength fluorescence excitation spectrum of MHN23. Column 1 and 2 contain the FWHM values, Dm, and the corresponding lifetimes, s, respectively. Column 3 shows the origin positions determined in step (b) of the procedure described in the text. Column 4 shows the merit parameter v2 for the best fits obtained; minimum is at Dm = 300 cm1.

normalised intensity

1.0

(a) (b)

(c)

0.5

0.2

0.2

-0.2

0.2

-0.2

(a)

26000

26500

27000

0.0

s = (2pcDm)1 (fs)

Origin (cm1)

v2 ¼ 1n

100 150 200 250 300 350 400 450 500

53.1 35.4 26.5 21.2 17.7 15.2 13.3 11.8 10.6

25870 25930 25950 25970 25980 26000 26010 26030 26070

0.0084(b) 0.0032 0.0028 0.0022 0.0019(a) 0.0021 0.0026 0.0027 0.0028(c)

(a) (b) (c)

(b)

0.0

25500

0.0

residuals

residuals

(c)

residuals

0.0

-0.2 27500

-1

wavenumber [cm ] Fig. 5. Illustration of the vibrational line fitting procedure for the long wavelength fluorescence excitation spectrum of MHN23. Top panel: Experimental data (d) and three fit curves (cf. Table 2). Bottom panel: residuals (Ei  Ti). Fit parameters: (a) Dm = 300 cm1, v2 = 0.0019 (best fit). (b) Dm = 100 cm1, v2 = 0.0084 (too many spectral features). (c) Dm = 500 cm1, v2 = 0.0028 (does not produce the resolved shoulder at 26100 cm1).

inadequate cooling of the MHN23 molecules, but rather results from the short lifetimes of the excited N tautomer states. A lower limit of the state lifetimes, s, can be estimated, by constructing the excitation spectrum based on the anticipated vibronic structure of the N tautomer, i.e., the positions, intensities and widths of the transitions involved. For that purpose the anticipated vibronic states are represented by Lorentzian line profiles with an effective FWHM, Dm, because the natural linewidths of the MHN23 rotational states are expected to greatly exceed the corresponding Doppler widths in the jet (0.09 ± 0.03 cm1) [1]. In Section 2.1 it was noted that the excitation spectra of the HB and NHB-1 conformers are very similar. Hence the N tautomer is expected to possess a similar vibrational structure as its rotamer equivalents. The measured line positions and intensities in the HB excitation spectrum (Table 1) are thus used as initial estimates of the parameters of the individual N tautomer vibronic transitions. The effective FWHM, Dm, of the N tautomer vibronic lines was estimated using the following procedure.

P ðEi T i Þ2

Dm (cm1)

Ei

Fit results shown in Fig. 5.

parameter, v2 = n1R(Ei  Ti)2/Ei where Ei and Ti are the experimentally measured and theoretically generated data points, and n is the number of fitted points. (c) The origin was then set at the value obtained in step (b) and the ‘Peak Fit’ function of Origin 8.01 was used to find the optimal fit to the experimental data, by allowing the intensities, Ak, of the individual Lorentzian profiles to vary by ±25% of the initial estimates. The position (xc) of each line profile and the FWHM of the profiles were fixed at the predefined values, outlined above. (d) Steps (a)–(c) were repeated for a series of Dm values from 100 to 500 cm1 in steps of 50 cm1. The fit quality was assessed using the v2 parameter defined above. The best fit to the experimental data (i.e., lowest v2 value) was obtained with a vibrational state line width of Dm = 300 cm1 (see Fig. 5). Table 2 shows the results of the fitting procedures for all values of Dm, while Fig. 5 shows three of the computed fits. The calculated parameters of the individual vibrational line profiles (xc, A) with Dm = 300 cm1 are also given in the Supplementary material (Table S2). A number of the relative intensity values decreased significantly from the initial estimates (e.g., xc = 172, 288, 313, 356, 432, and 439 cm1 all adopted the minimum allowed intensity value A). This may indicate that these features have negligible intensity in the N tautomer excitation spectrum. As Dm was varied in steps of 50 cm1, a minimum error of ±25 cm1 should be attached to the computed best fit vibrational line width value, i.e., Dm = 300 ± 25 cm1. Hence, a lower limit can be placed on the average N tautomer rotational state lifetime, of s P 18 ± 2 fs. 2.3. Intramolecular dynamics and proton transfer (pt)

(a) The widths of all the Lorentzian line profiles were fixed at some uniform value, Dm. (b) A series of Lorentzian line profiles were constructed with energies (xc) (see Supplementary material, Table S2), relative to the origin, defined by the excitation spectrum of the HB conformer (Table 1). An initial guess of the absolute intensity (A0) of the origin line was made and the intensities of all other Lorentzian lines, relative to this value were fixed at the relative intensity values measured in the HB excitation spectrum. The energy of the origin of the S1 S0 electronic transition was then estimated by varying the origin position in steps of 10 cm1 (i.e., shifting the resultant profiles along the energy axis), in order to obtain the best possible fit to the experimental data. In each case the quality of the fit was determined by calculating the merit

The time-dependence of the short-wavelength fluorescence of MHN23 at room temperature has been reported to be biexponential [11]. The corresponding lifetimes of 14.9 ns and 2.9 ns, established with a time resolution of 30 ps, were tentatively assigned to the HB conformer and to one of the NHB conformers, respectively, which is in agreement with our assignment of the dominant features in the vibronic structure of the fluorescence excitation spectrum. Spectra obtained for MHN23 in solution [10,11] and in a static gas [11] showed no evidence of prompt fluorescence from the N tautomer (as mentioned in the introduction). However, since the potential occurrence of a dual N tautomer fluorescence cannot 1 The Peak Fit function of Origin 8.0 performs non-linear least square fitting procedures of multiple peaks in a single dataset.

A. McCarthy, A.A. Ruth / Chemical Physics 425 (2013) 177–184

be ruled out per se, the observation of the ‘‘humped’’ structure in Fig. 3 and the possible explanation (iii) in Section 2.2 will be reconsidered briefly after the following discussion of the intramolecular dynamics of the N tautomer: In Ref. [11] Catalan et al. arrived at the conclusion in conjunction with data from Law and Shoham [10], that only a fraction of excited molecules undergo ESIPT with a yield of Upt  0.018 (cf. Fig. 2). The quantum yield of prompt fluorescence, Ur, was estimated to be 0.003, and it was concluded that the N tautomer thus decays predominantly via an efficient non-radiative mechanism. Based on this information and the lower limit of s (P18 fs) established in the previous section, the upper limit of the rate coefficient for ESIPT, kpt, can be estimated as:

kpt ¼ ðkpt þ knr þ kr ÞUpt  s1 ð0:018Þ 6 1  1012 s1

ð1Þ

where knr and kr are the rate cofficients for all non-radiative and radiative processes, respectively. The proton transfer lifetime spt P 1 ps as per Eq. (1), would consistent with the presence of an energy barrier to ESIPT on the S1 potential energy surface. DFT B3LYP/ 6-31G** and CIS 6-31G** calculations of the potential energy surfaces of MHN23 [11] predicted a 1.5 kcal/mol ( 525 cm1) energy barrier between the N tautomer and keto tautomers on the S1 potential energy surface. Assuming a typical rate coefficient for radiative decay, kr  108  109 s1  kpt, the rate of the non-radiative process can be estimated as knr = s1  (kpt + kr)  (s1  kpt) 6 5.5  1013 = s1  (kpt + kr)  (s1  kpt) 6 5.5  1013 s1. Catalan et al. [11] proposed intersystem crossing as a possible efficient non-radiative channel in the N tautomer. Although rare, intersystem crossing rates of this magnitude in organic molecules (i.e., kisc  1013 s1) have been reported in the literature[15-19]. It is not immediately obvious, however, what mechanism would promote efficient ISC in MHN23, since spin–orbit coupling would not be expected to be particularly strong [20]. Moreover, a luminescence which could be attributed to phosphorescence was not evident in the experiment. Ultrafast S1 ? S0 internal conversion appears unlikely as the dominant non-radiative process, based on the potential energy surface calculation, in Ref. [11] where the minimum separation of the S0 and S1 surfaces was estimated as 80 kcal/mol (27980 cm1). However, fast S1 ? S0 internal conversion could be relevant if a conical intercept along a reaction coordinate favours non-radiative decay. Indeed, based on the emission spectrum in Ref. [11] there is a substantial Stokes-shift of the short-wavelength emission which indicates a significant displacement of the upper state potential surface relative to the ground state of the emitting species (assumed to be not N*). Given the shallow S1 excited state potential hypersurface, there is a reasonable probability for an S0–S1 conical intersection of the N species. Azulene is a well-known example, with an S1 lifetime of about 1 ps determined by conical intersection via displacement along the trans-annual bond [21]. Radiationless decay of N* promoted by displacement along an O–H coordinate with tunneling (but in a region of the surface where net proton transfer does not occur) could be even significantly faster, as required if the rate constant for proton transfer is <1012 s1 and the ESIPT yield is about 2%. Furthermore, if one accepts that the short wavelength emission has a quantum yield of 0.003, then the quantum yield of long wavelength emission can be assumed to be about an order of magnitude smaller, based on the spectra in [11]. The quantum efficiency of radiative decay of the PT tautomer then becomes (0.1  Ur)/Upt  0.017. Catalan et al. report a lifetime of 60 ps for the long wavelength emission in cyclohexane solution at room temperature, which yields reasonable (i.e. consistent) radiative and non-radiative decay rate coefficient for the PT tautomer of kr = 3  108 s1 and knr = 1.6  1010 s1, respectively. On the observability of the prompt fluorescence of the N tautomer

183

Arguments that would support (Pro) or oppose (Con) an assignment of the broadband underlying structure in Fig. 3 as prompt fluorescence from the N tautomer are as follows: (Pro 1) Of all conformers the N tautomer ground state is the most populated before excitation. The relative conformer populations are retained upon expansion in the jet. (Pro 2) In an isolated cold state in a supersonic expansion the probability for radiative decay (rate coefficient, kr) is increased in comparison to MHN23 in solution at room temperature or in a static gas. (Pro 3) Inhomogeneous broadening in solution (or in a static gas) makes it virtually impossible to spectrally distinguish species with dramatically different decay dynamics. If the ‘humped’ structure in the short-wavelength excitation spectrum in Fig. 3 were due to prompt fluorescence from the excited N tautomer, a lower limit of the lifetime could be estimated by fitting a Lorentzian profile to the lowest broad feature. Such a fit, shown in Fig. 3, yields a FWHM of 57 cm1 corresponding to a lower limit lifetime of 93 fs, which is on the same order of magnitude as the lifetime s P 18 fs estimated from modelling the long wavelength excitation spectrum (Section 2.2). Fluorescence from a state with such a short lifetime could not be observed in the time-dependent measurements in Refs. [10] and [11] owing to the limit in time resolution. (Pro 4) If the ‘humped’ structure were not due to any of the two alternative explanations (i) and (ii) in Section 2.2, then there is no other obvious other species or product that could cause the observed emission. (Con 1) The estimated rate coefficients and corresponding quantum yield would make the fluorescence impossible to observe. Assuming a yield of 3  103 and the assumption that the 93 fs lifetime is dominated by non-radiative decay, kr would have to be 3  1010 s1, which appears 2–3 orders of magnitude too large. This is a rather strong argument, since it is unlikely that the intramolecular dynamics changes dramatically by going from a room temperature system in solution or a static gas to a jet-cooled sample. (Con 2) The onset of the potential prompt fluorescence from the N tautomer is blue-shifted with respect to the long-wavelength excitation spectrum, and the fact that the room temperature absorption spectra reported in Refs. [10,11] resemble the longwave excitation spectrum (Section 2.2). If the broad structure were prompt N tautomer fluorescence the non-radiative decay would have to be strongly excitation energy dependent at low excess energies over the S1,0 origin. A quantitative explanation of the blue-shift with respect to the onset of the long-wavelength emission can however not be given on basis of the new data presented here. However, in this context it may be important to note that a number of the totally symmetric vibrations, which feature prominently in the short-wavelength excitation spectrum, may promote proton transfer in the N tautomer. In particular, the C–C-angle bend vibrations m010 ; m011 ; m014 ; m018 and m020 all strongly modulate the O20  H18 distance in the N tautomer. Therefore proton transfer may be very efficient below 650 cm1 above the origin and the corresponding yield might be based on these ‘‘gateway states’’ in the corresponding energy region.

3. Conclusions Two distinct excitation spectra have been observed for jetcooled MHN23, by monitoring the short and long wavelength emission bands separately. The excitation spectrum of the short wavelength fluorescence exhibits a resolved vibrational structure while the excitation spectrum of the long wavelength fluorescence exhibits a broad, rather unstructured band. The observation of

184

A. McCarthy, A.A. Ruth / Chemical Physics 425 (2013) 177–184

these two distinct spectra, confirms that the emission bands do not arise from the same absorbing species. As the long wavelength emission is assigned to the proton transfer tautomer after ESIPT of the excited N tautomer, the short wavelength emission were assigned to the HB and NHB conformers, as previously ascertained by Catalan et al. [11]. At least two species contribute to the short wavelength excitation spectrum. The dominant species (S0 ? S1 origin at 26743.7 cm1) is assigned as the HB conformer. A series of weaker lines with an apparent S0 ? S1 origin line at 26372.7 cm1 have been assigned to the NHB-1 conformer. This result corroborates and further specifies the finding by Catalan et al. [11] who tentatively assigned a portion of the short wavelength emission to a NHB conformer. As the energy separation of the NHB-1 and NHB2 conformers is small the latter conformer may also be observed in the excitation spectrum. A weak feature at 26356 cm1 has been tentatively assigned as the origin of the S0 ? S1 electronic transition of the NHB-2 conformer. A broad underlying ‘‘humped’’ background signal could potentially be due to prompt fluorescence from the N tautomer and arguments for and against such an assignment have been outlined. The broad long wavelength excitation spectrum was satisfactorily modelled with a set of Lorentzian lines with a FWHM of (300 ± 25) cm1. To aid the fitting procedure, the energies of the lines and their relative intensities were assumed to closely resemble the excitation spectra of the HB and NHB-1 conformers. The N tautomer in the S1 state decays predominantly via an ultrafast nonradiative process with a rotational state lifetime that was estimated as, s P 18 ± 2 fs (knr 6 5.5  1013 s1). Assuming a proton transfer efficiency of 1.8% [10,11], the proton transfer lifetime was calculated as, spt P 1 ps. If the prompt N tautomer fluorescence were indeed observed in the spectrum, then spt may be as large as 5.3 ps. Both values are consistent with the presence of an energy barrier to ESIPT, as predicted by the calculations of Catalan et al. [11]. The ultrafast nature of the non-radiative decay could not be established based on the data presented.

Acknowledgment AMcC acknowledges receipt of a studentship through the Irish Research Council for Science Engineering and Technology (IRCSET), EMBARK initiative 2007. Fruitful discussions with Professor Paul Brint are also gratefully acknowledged. Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.chemphys.2013. 08.017. References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21]

A. McCarthy, A.A. Ruth, Phys. Chem. Chem. Phys. 13 (2011) 7485. A. McCarthy, A.A. Ruth, Phys. Chem. Chem. Phys. 13 (2011) 18661. J.L. Herek, S. Pedersen, L. Banares, A.H. Zewail, J. Chem. Phys. 97 (1992) 9046. L. Heimbrook, J.E. Kenny, B.E. Kohler, G.W. Scott, J. Phys. Chem. 87 (1983) 280. L. Heimbrook, J.E. Kenny, B.E. Kohler, G.W. Scott, J. Chem. Phys. 75 (1981) 5201. R. Lopez-Delgado, S. Lazare, J. Phys. Chem. 85 (1981) 763. E.D. Bergmann, Y. Hirshberg, S.J. Pinchas, Chem. Soc. 482 (1950) 2351. Y.V. Naboikin, B.A. Zadorozhny, E.N. Pavlova, Optika Spektrosk. 6 (1959) 492 (Engl. Transl. Opt. Spectrosc. 6, 312). G. Woolfe, P. Thistlethwaite, J. Am. Chem. Soc. 103 (1981) 3849. K.-Y. Law, J. Shoham, J. Phys. Chem. 98 (1994) 3114. J. Catalan, J.C. del Valle, J. Palomar, C. Diaz, J.L.G. de Paz, J. Phys. Chem. A 103 (1999) 10921. S. Beck, D. Powers, J. Hopkins, R. Smalley, J. Chem. Phys. 73 (1980) 2019. A.A. Ruth, W.G. Doherty, R.P. Brint, Chem. Phys. Lett. 352 (2002) 191. R.S. Ruoff, T.D. Klots, T. Emilsson, H.S. Gutowsky, J. Chem. Phys. 9 (3) (1990) 3142. J. Jortner, Philos. Trans. Roy. Soc. Lond., Ser. A 356 (1998) 477. Y. Ohshima, T. Fujii, T. Fujita, D. Inaba, M. Baba, J. Phys. Chem. A 107 (2003) 8851. H.J. van Ramesdonk, B.H. Bakker, M.M. Groeneveld, J.W. Verhoeven, B.D. Allen, J.P. Rostron, A. Harriman, J. Phys. Chem. A 110 (2006) 13145. A. Cannizzo, A. Blanco-Rodriguez, A. El Nahhas, J. Sebera, S. Zalis, A. Vlcek, M. Chergui, J. Am. Chem. Soc. 130 (2008) 8967. G.J. Hedley, A. Ruseckas, I.D.W. Samuel, J. Phys. Chem. A 113 (2009) 2. N.J. Turro, Modern Molecular Photochemistry, University Science Books, Sausalito CA, 1991, p. 48 (Chapter 6.4). A.A. Ruth, E.-K. Kim, A. Hese, Phys. Chem. Chem. Phys. 22 (1999) 5121.