Rotationally resolved UV spectroscopy of the rotamers of indole-4-carboxylic acid: Evidence for charge transfer quenching

Abstract

Rotationally resolved electronic spectra of two conformational isomers of jet-cooled indole-4-carboxylic acid (I4CA) and the deuterated forms of the acid (—COOD) and amide (—ND) groups have been obtained using a UV laser/molecular beam spectrometer. The in-plane orientation of the acid group defines the two lowest energy rotamers of I4CA. The S₁ ← S₀ origin bands of the two rotamers and four isotopologues have been fit to asymmetric rotor Hamiltonians in both electronic states. From the best-fit parameters, the positions of the H-atoms in the principal axis frames of each conformer have been determined and serve to unambiguously identify the syn forms (i.e., COH…O) of the cis and trans rotamers. The experimental S₀ and S₁ inertial parameters, hydrogen atom positions, and transition dipole moment (TDM) orientations are compared with the results of theoretical calculations. The TDM orientation indicates that the S₁ state is the ¹L_a state in contrast to most substituted indoles. The molecular orbital properties and natural charges are investigated to better understand the ¹L_a/¹L_b state reversal and the extent of photoinduced intramolecular charge transfer that impacts the rotamer-dependent fluorescence lifetimes.

I. INTRODUCTION

Amino acids often present interesting photophysical and spectroscopic properties, owing to their unique natural composition and structure. The amino acid, tryptophan, and the neurotransmitters generated from it, tryptamine, serotonin, and melatonin, are all derivatives of indole, which are known to be associated with important biological reactions in living cells and brain function.¹ The indole-based carboxylic acids are also important biochemically active reagents. Given the importance of the carbonyl group in biochemical interactions (e.g., protein–protein interactions), various isomeric forms have been the subject of numerous spectroscopic studies in the gas^2–4 and condensed phases.^3,5 These studies were aimed in part at elucidating the dependence of the fluorescence decay properties and photoinduced changes in the acid/base properties of the chromophore on the substituent position and functionality.

Following the nomenclature of Platt⁶ for aromatic hydrocarbons, the classification of the ¹L_a and ¹L_b states of indole has also been used to qualitatively understand the excited state properties of indole and its derivatives.⁷ The ¹L_a electronic state is largely composed of the LUMO (lowest unoccupied molecular orbital) ← HOMO (highest occupied molecular orbital) transition, while the ¹L_b state is principally composed of two transitions, LUMO ← HOMO-1 and LUMO+1 ← HOMO.⁸

Condensed phase studies of the 4- and 5-carboxylic acid isomers have shown that fluorescence can be observed from both the ¹L_a and ¹L_b states, which are closely spaced.¹ Upon photoexcitation in polar solvents, the ¹L_a state is stabilized to a greater extent than the ¹L_b state because of the larger dipolar character of the former. In the gas phase, the results from jet-cooled fluorescence excitation studies of the 3-, 4-, and 5- acid derivatives have shown a position dependence of the ¹L_a/¹L_b state order.⁴ The S₁ states of the 4- and 5-acid isomers have been assigned to ¹L_b states, which is suggested to result from electronic conjugation effects between the acid group and the benzene ring moiety.⁴ However, at the 3-position, conjugation with the pyrrole ring is not feasible and has been argued to explain observed features that characterize the ¹L_a state.

These studies were based on observed progressions, or lack of them, in the excitation spectra, or the appearance of red-shifted features in the dispersed emission spectra. More direct evidence comes from the determination of the transition dipole moment (TDM) orientation from the rotationally resolved UV studies in a molecular beam. Berden and co-workers⁹ have investigated the S₁ ← S₀ band origin of indole and found that the TDM orientation is +38.3°, in line with the expected polarization of the ¹L_b state (see Fig. 1).^10,11 However, a similar study of 4-cyanoindole by Hebestreit and co-workers¹² shows that the TDM orientation and a permanent dipole moment increase in S₁ are indicative of the ¹L_a state, in agreement with their singles and doubles coupled cluster calculations. A similar study of the different conformers of 4-, 5-, and 6-methoxyindole¹³ was found to introduce some ¹L_a state character into the S₁ states of the 4- and 6-isomers. Additional results from studies of the S₁ and S₂ states of 4- and 6-fluoroindole indicated the ¹L_a/¹L_b character to be highly mixed in the 6-isomer, while the S₁ state of the 4-isomer was assigned to ¹L_b based on the observed change in the permanent dipole moment.¹⁴

FIG. 1.

The two syn (s) conformations of I4CA and indole shown in the principal axis frames. The atom numbering and the observed TDM are shown for indole.

The conjugation between the acid group and the aromatic ring would be expected to enhance stabilization of the ¹L_b state relative to ¹L_a via an excited state charge transfer (CT) mechanism. Indeed, in the previous study of I4CA,² Sulkes et al. observed two origin bands in its low-resolution UV spectrum; the lower energy band (band A) has a shorter lifetime and was assigned to the s-cis conformer because of the more favorable orientation of the carboxyl group for CT quenching. (The two lowest energy conformers of I4CA are shown in Fig. 1.) The higher energy band, band B, exhibits a >50% longer lifetime, suggesting that the orientation of the carbonyl group does have a significant impact on the quenching mechanism. Other studies have shown that other substituted indoles exhibit nonexponential fluorescence decay,^4,15,16 a fact that has also been attributed to intramolecular electron transfer.¹⁷

In order to further investigate the conformationally dependent excited state properties of I4CA, we have recorded the rotationally resolved S₁ ← S₀ fluorescence excitation spectra of the two conformers and the four deuterium isotopologues of —COOH and —NH groups in a molecular beam. From these data, we determine the H-atom substitution coordinates of these groups to unambiguously assign the origin bands of the two conformers. We also obtain for the first time the TDM orientations which report on the ¹L_b and ¹L_a state character of the S₁ states. The rotational constants, substitution coordinates, and TDM orientations are compared with predictions from ab initio calculations of the ground and excited states to provide additional information about the conformer geometries, the electronic state character, and the degree of intramolecular charge transfer that impacts the fluorescence lifetimes of the two rotamers.

II. EXPERIMENTAL

I4CA is available commercially (AA Blocks, LLC) and was used without further purification. Deuterated samples were prepared after one H/D exchange by mixing 5 gm of sample with 100 ml of D₂O. After sitting for two days, the solid sample was separated using a 4 μm pore size filter paper. Fluorescence excitation spectra of the S₁ ← S₀ transitions of I4CA were measured using a UV laser/molecular beam spectrometer described elsewhere.¹⁸ Briefly, a continuous-wave ring dye laser (Coherent 699) operating with a dye mix of Kiton Red (Exciton, Inc.) and DCM special (Exciton, Inc. [2-[2-[4-(dimethylamino)phenyl]ethenyl]-6-methyl-4H-pyran-4-ylidene]-propanedinitrile) was pumped with 4 W from an Ar⁺ laser (Spectra 171, 514 nm line) to generate 120–200 mW of narrowband light (≈1 MHz) near 640 nm. The dye laser was mode-matched to an externally resonant cavity (WaveTrain I, Spectra-Physics, Inc.) containing a β-barium borate crystal (Conex Systems Technology, Inc.) and generated 2–4 mW of the UV light near 320 nm. I4CA was heated to 190 °C in a three-chamber quartz source and the vapor was mixed with 29 kPa (220 torr) of Ar gas and expanded into a source chamber through a 150 μm diameter tapered quartz nozzle tip. The molecular beam was skimmed and then crossed by a mildly focused UV beam about 18 cm downstream of the source. The fluorescence at the beam crossing was collected at 90° to both beams with ≈20% collection efficiency using two spherical mirrors¹⁹ and detected using a photomultiplier (Thorn EMI, 9813QB) and computer-interfaced photon counter (SSR, Inc, 1110). The Doppler limited resolution of 21.5(10) MHz was determined directly from line shape measurements of 1-fluoronaphthalene (Aldrich, Inc) at a source temperature of 100 °C since the natural linewidth contribution is relatively minor, i.e., 1.46 MHz at 320 nm.²⁰

The electronic jet-cooled spectra of I4CA were obtained at a source temperature of 190 °C. The photon count rate of the strongest lines of the parent species was in excess of several million counts per sec (CPS). For our discriminator (PAR, Inc., 1121) dead time, these high rates are significantly underestimated because of pulse pile-up error, which tends to suppress and, therefore, decrease the intensity of the strongest a-type lines (ΔJ = 0, ±1 ΔK_a = 0, K_c = ±1) relative to the weaker b-type lines (ΔJ = 0, ±1 ΔK_a = ±1, K_c = ±1). To bring the count rate into a linear response limit for accurate determination of hybrid band ratios, an aperture in front of the PMT at the fluorescence focal point was reduced to keep the maximum count rate at less than 150 kCPS.

The frequency tuning of the dye laser was performed using a scanning acousto-optic modulator (AOM) sideband system whose double-diffracted output was continuously locked to a fringe of an evacuated HeNe stabilized reference cavity.^18,21 Rapid jumps of the AOM frequency to lock to consecutive fringes enabled continuous tuning across the dye laser single-mode tuning range of 0.75 cm⁻¹ (1.5 cm⁻¹ in the UV). Typically, fluorescence spectra were obtained in two or three contiguous 0.75 cm⁻¹ sections with each section consisting of the signal averaged data of three repeated scans. The photon counter data were first corrected for residual saturation effects, normalized to UV power, and then linearized based on the radio frequency calibrated AOM ramp. The repeated scans and overlapped sections were spliced together at integral fringe spacings and averaged using automated features of the JB64 software.^22–24 The free spectral range (FSR) of the reference cavity was determined from numerous selected measurements of the I₂ spectrum²⁵ covering a 385 cm⁻¹ region (>23 000 fringes) that spanned the two origin regions of I4CA. The calculated FSR is 500.485(3) MHz, where the uncertainty (type A, k = 1 or 1σ)²⁶ was determined from propagation of the reported uncertainties of ±60 MHz for the I₂ lines used. It is noted that the absolute frequency positions of the fringes measured for all bands were maintained over the entire measurement period. Absolute frequencies are accurate to ±0.002 cm⁻¹ and were determined using a wavemeter (Toptica Photonics, Inc., HighFinesse WS-7) sampling at 150 Hz after application of a single correction factor based on the average difference relative to several corrected I₂ absorption lines.²⁵

The spectral analysis was performed using the JB64 suite program.^22–24 The rotational Hamiltonian used to fit the spectra was an asymmetric-top rigid-rotor model. The rotationally resolved spectra were initially fit and then later refined using genetic algorithms (GAs),^27,28 beginning with rough estimates of ground state rotational constants from ab initio theory.²⁹ A cluster version of the GA program was used to concurrently execute 10 parallel and independent runs, which were completed in <10 min. Each run began with a randomly seeded population for each parameter. The small inertial defects observed (vide infra) for all states indicate nearly planar structures and, therefore, reduced the number of parameters needed to fit the line intensities. The intensity parameters included the Lorentzian component of the Voigt line shape function; three parameters of a two-temperature model;^27,30 one angle to define the in-plane TDM orientation, θ_TDM; and a second angle to define the excited-state in-plane inertial-axis-reorientation angle, θ_a/b. All GA runs were tightly converged after 500 generations. The spectral simulations for each solution were nearly indistinguishable when overlaid with the experimental data. Automated line assignments were made based on each GA run at the digital resolution of 1 MHz and were used in linear least squares analyses to determine the best fit rotational constants. Standard deviations of the parameter differences across all runs were compared with the quadrature sums of the standard deviations from the least-squares analyses. These two measures of uncertainties were almost always within a factor of two of one another. Furthermore, additional GA runs were performed following changes in the FSR by one standard deviation. The largest uncertainties of the three methods are reported in Tables I and II. The uncertainties of the rotational constants are shown for the least significant figures and are type A, k = 1 or 1 σ. The parameters related to the intensity parameters are determined from the standard deviations over numerous GA parameter sets which included changes to the baseline and line shape models and are, therefore, type B, k = 1 or 1 σ.

TABLE I.

Best fit rotational constants determined for conformer A of I4CA and the COOD and ND isotopologues in their ground (S₀) and excited (S₁) electronic states. Uncertainties of the rotational constants are specified for the least significant digits from linear least squares fits and are type A, k = 1 or 1 σ. Other uncertainties obtained from genetic algorithms are type B, k = 1 or 1 σ (see text for details).

	Band A
Parameter	Parent	d1 (—ND)	d1 (—COOD)
A″ (MHz)	1 613.162(40)	1 612.920(80)	1 612.433(43)
B″ (MHz)	866.595(4)	847.124(7)	844.487(7)
C″ (MHz)	564.085(2)	555.768(7)	554.611(3)
ΔI (amu Å2)a	−0.53(1)	−0.58(2)	−0.64(2)
ΔA (MHz)	−11.385(1)	−11.607(2)	−11.467(2)
ΔB (MHz)	−6.181(1)	−5.981(2)	−5.843(1)
ΔC (MHz)	−4.119(1)	−4.066(2)	−3.984(1)
ΔΔI (amu Å2)	+0.175(2)	+0.189(2)	+0.179(2)
Origin (cm−1)	31 122.781(2)	31 111.965(2)	31 130.098(2)
Δv (cm−1)	0.0	−10.816(2)	+7.317(2)
Inertial frame, θa/bb	−0.96(2)	−0.96c	−0.96c
Ia/Ib (%)	93.2(10)/6.7(10)	92.3(10)/7.7(10)	93.9(10)/6.1(10)
θTDMd	±15.1(10)	±16.1(11)	±14.3(11)
Temp. (T1/T2/wt.)e	3.4/9.1/0.38	4.4/14/0.53	3.2/9.7/0.51
Loren. FWHMf (MHz)	43(2)	48(2)	48(3)
Fit lines	462	389	320
OMC (MHz)g	1.1	2.2	2.1

^aΔI = 505 379(1/C-1/A-1/B), where A = ħ²/2I_a, where I_a is the moment of inertia about the a-axis in amu Ų.

^bAxis reorientation angle in the a/b inertial plane where negative values correspond to the clockwise rotation of the S₁ frame relative to S₀.

^cFixed in fit to the parent values because of the reduced signal-to-noise level.

^dθ_TDM = ±atan(I_b/I_a)^0.5.

^eTwo-temperature model from Refs. 8 and 26: exp(−ΔE/k_bT₁) + wt exp(−ΔE/k_bT₂), where ΔE is the ground state energy and k_b is the Boltzmann constant.

^fFWHM is full width at half maximum, gaussian component fixed at 21.5(10).

^gObserved minus calculated standard deviation.

TABLE II.

Best fit rotational constants determined for conformer B of I4CA and the COOD and ND isotopologues in their ground (S₀) and excited (S₁) electronic states. Uncertainties of the rotational constants are specified for the least significant digits from linear least squares fits and are type A, k = 1. Other uncertainties obtained from genetic algorithms are type B, k = 1 (see text for details).

	Band B
Parameter	Parent	d1 (—ND)	d1 (—COOD)
A″ (MHz)	1 590.982(31)	1 590.256(140)	1 572.786(43)
B″ (MHz)	876.787(7)	856.671(31)	863.046(15)
C″ (MHz)	565.686(5)	557.180(66)	557.664(5)
ΔI (amu Å2)a	−0.66(2)	−0.70(11)	−0.66(4)
ΔA (MHz)	−8.269(1)	−8.511(8)	−7.739(5)
ΔB (MHz)	−10.402(2)	−10.003(5)	−10.172(3)
ΔC (MHz)	−5.524(1)	−5.433(5)	−5.363(2)
ΔΔI (amu Å2)	+0.230(2)	+0.252(6)	+0.226(3)
Origin (cm−1)	31 505.556(2)	31 495.751(4)	31 514.482(2)
Δv (cm−1)	0.0	−9.805(4)	+8.926(2)
Inertial frame θa/bb	−0.31(1)	−0.31c	−0.31d
Ia/Ib (%)	87.3(10)/12.7(10)	87.3/12.7c	87.3/12.7d
θTDMe	±20.9(10)	±20.9c	±20.9d
Temp. (T1/T2/wt.)f	3.8/10/0.23	4.7/14/0.50	4.0/14/0.62
Loren. FWHMg (MHz)	27(2)	27c	27d
Fit lines	467	247	320
OMC (MHz)h	1.2	2.5	2.1

^aΔI = 505 379(1/C-1/A-1/B), where A = ħ²/2I_a, where I_a is the moment of inertia about the a-axis in amu Ų.

^bAxis reorientation angle in the a/b inertial plane where negative values correspond to the clockwise rotation of the S₁ frame relative to S₀.

^cFixed in fit to the parent values because of sever overlap with the vibronic band of A at v_A + 383 cm⁻¹.

^dFixed in fit to the parent values because of an insufficient signal-to-noise level.

^eθ_TDM = ±atan(I_b/I_a)^0.5.

^fTwo-temperature model from Refs. 9 and 30: exp(−ΔE/k_bT₁) + wt exp(−ΔE/k_bT₂), where ΔE is the ground state energy and k_b is the Boltzmann constant.

^gFWHM is full width at half maximum, gaussian component fixed at 21.5(10) MHz.

^hObserved minus calculated standard deviation.

Theoretical calculations were performed using the G16/avx version of the Gaussian program suite.²⁹ Ground state geometry optimizations were done at the density functional theory (DFT) level using the functional B3LYP³¹ and a triple zeta basis set that includes the Grimme dispersion correction (Gaussian keyword: Def2TZVP empiricaldispersion = gd3bj).³² Excited state geometry optimizations were performed using time dependent DFT (TD-DFT)³³ using the functional B3LYP, with the same basis set. For the natural bond orbital (NBO) analysis,³⁴ the NBO archive file (version 3.1) generated by the Gaussian program was used as input to GenNBO³⁴ (version 5.1) to calculate natural charges and orbitals.

III. RESULTS

A. Parent molecules

The S₁–S₀ fluorescence excitation spectrum of band A at 31 122.8 cm⁻¹ (321.31 nm) is shown in the top panel of Fig. 2, and the corresponding spectrum of band B at 31 505.6 cm⁻¹ is shown in Fig. 3. Both spectra span about 2.5 cm⁻¹ and exhibit well-defined P, Q, and R branches. The origin splitting is 382.8 cm⁻¹. An expanded portion of each band is shown in the lower panels, and the best fit simulated spectra are superimposed. As evident from the residuals in each panel, the bands are fit to nearly within the signal-to-noise level which exceeds 300:1 for these two bands. Including first-order distortion terms in either state did not improve the quality of the fits.

FIG. 2.

Rotationally resolved S₁–S₀ fluorescence excitation spectrum of band A of I4CA (upper panel) and an expanded portion of the spectrum (lower panel). The experimental data are shown as black lines, and the simulated spectra are shown with (blue lines) and without (red lines) a convoluted line shape function. The unscaled residuals are shown as the lower trace in each panel.

FIG. 3.

Rotationally resolved S₁–S₀ fluorescence excitation spectrum of band B of I4CA (upper panel) and an expanded portion of the spectrum (lower panel). The experimental data are shown as black lines and the simulated spectra are shown with (blue lines) and without (red lines) a convoluted line shape function. The unscaled residuals are shown as the lower trace in each panel.

The rotational constants from the fits are listed in the first column of Tables I and II for bands A and B, respectively. The ground state inertial defects, ΔI, are small as expected for nearly planar structures and negative, indicating larger contributions from out-of-plane vs in-plane zero-point vibrational motions.³⁵ We also note that the inertial defects are more than an order-of-magnitude larger than those recently reported for 4-cyanoindole.¹² The rotational constants of the excited states are specified as changes relative to S₀ and in all cases are all smaller than their ground state counterparts. The S₁ inertial defects, also specified as changes, are both less negative relative to S₀. We also note that depending on the absolute TDM signs which are not determined in this analysis, the TDM angles, θ_TDM, of the two bands differ by as little as 6° or as much as 38°.

The Lorentzian linewidths observed for the protonated bands of A and B are 43(2) MHz and 27(2) MHz and correspond to lifetimes of 3.7(3) ns and 6.0(3) ns, respectively. These lifetimes are about 10% shorter than the previously reported values of 4.2(2) ns and 6.4(2) ns, respectively, determined from single photon measurements.² One possible explanation for the increased width of the frequency-resolved measurements is the ¹⁴N hyperfine structure. We have estimated the J dependent ¹⁴N splittings using first order perturbation theory as implemented in the JB64 software.

Simulated spectra (not shown) were generated using the hyperfine constants predicted for the S₀ and S₁ states at the DFT and TD-DFT levels of theory, respectively. In general, the hyperfine splitting only increased the widths by a few percent at most. A second possible reason for this increase in width may be the higher beam velocity of the I4CA measurements compared to 1-fluoronaphthalene (1FN) where the Gaussian width contribution to the Voigt profile was determined. The source temperature used for I4CA was about 100 °C higher than for 1FN. This is expected to increase in the Doppler width by ≈10% and, therefore, could easily account for these differences.

B. Deuterium isotopologues

The simple exchange procedure discussed above for the deuteration of the sample is expected to lead to exchange of only the labile protons at the —COOH and —NH sites of I4CA. (Exceptions to this rule have been observed before for 1-aminonaphthalene.³⁶) Furthermore, a partial exchange was targeted to obtain the singly substituted forms (d₁) needed for the substitution analysis described below. We have estimated based on the intensity ratios of the spectra that only about 3% exchange occurred, which proved to be sufficient given the high signal-to-noise ratios of the parent species. Although other deuterium substituted bands are present, only two are reported here for band A and are shown in Figs. S1 and S2 of the supplementary material. The best-fit rotational constants are given in the last two columns of Table I. The corresponding spectra of band B are shown in supplementary material, Figs. S3 and S4, and the parameters are included in Table II. The relative frequency shifts in Tables I and II locate them to the blue and red sides relative to the corresponding parent spectra by about (7–11) cm⁻¹. We also note that the d₁ band of B in Fig. S4 is partially overlapped with a strongly perturbed vibronic band of A. However, part of the central a-type Q-branch and R-branch are isolated enough to obtain reliable fits.

Because of the mass increase upon deuterium substitution, the S₀ rotational constants of the four deuterated forms are smaller than those of the parent species. While many but not all of S₀ and S₁ changes in the constants in Tables I and II follow this trend, the absolute S₁ constants are all smaller as well. We also note for the A bands that the inertial defects, TDM angles, and linewidths of the d₁ bands are similar to those of the corresponding parent species.

C. Substitution analysis

The rotational constants in Tables I and II give structural information about the inertial properties of the parent and deuterated species and can be used to determine accurate values of the center-of-mass coordinates of the substituted atom in the principal axis frame of the parent using Kraitchman’s equations.³⁷ As mentioned above, the small inertial defects indicate near-planar structures for all observed spectra, and therefore, only the in-plane coordinates along the a- and b-principal axes were calculated using a planar reduction of the equations. Uncertainties of the substitution coordinates are determined using Costain’s criteria³⁸ based on the rotational constant uncertainties.

For the A and B band regions, the in-plane coordinates of the two substituted atoms were obtained from the analysis of the parent’s rotational constants and those of the two nearby isotopologues. These are listed in Table III. (The signs of the coordinates are not determined from the analysis but are included by reference to the displayed structures in Fig. 1.) For the two red-shifted bands, the changes in magnitude upon electronic excitation undergo small increases, while those of the blue-shifted bands show both increases and decreases. We note that in many cases, the changes are smaller than the combined uncertainties. Furthermore, comparing the absolute values of the coordinates determined for the two red-shifted bands, the differences are less than 0.2 Å in both states, indicating that this atom is not likely associated with the coordinate change expected for the cis and trans rotamers.

TABLE III.

Experimental substitution coordinates and center-of-mass positions of two hydrogen atoms in the S₀ and S₁ states from the analysis of bands in the A and B regions of I4CA. The predicted H-atom coordinates of the —COOH and —NH groups are given in both electronic states for the s-cis and s-trans conformers of I4CA. All coordinates are specified relative to the ab-principal axis frames. Coordinate signs are included for consistency with displayed structures and were not determined from the experimental data.

	Band A region			s-cis form
	S0	S1	Δr (S1–S0)	S0	S1	Δr (S1–S0)
		—ND		DFTa	TD-DFTb
a (Å)	−3.655(3)	−3.664(3)	−0.009	−3.669	−3.662	+0.007
b (Å)	+0.222(42)	+0.310(30)	+0.088	+0.218	+0.291	+0.073
R (Å)	3.662(4)	3.677(4)	+0.015	3.675	3.674	−0.001
		—COOD
a (Å)	+3.893(2)	+3.891(2)	−0.002	+3.910	+3.902	−0.008
b (Å)	+0.387(16)	+0.411(15)	+0.024	+0.313	+0.344	+0.031
R (Å)	3.912(3)	3.913(3)	+0.001	3.922	3.917	−0.005
	Band B region			s-trans form
		—ND
a (Å)	−3.663(14)	−3.672(14)	−0.010	−3.652	−3.646	+0.006
b (Å)	+0.376(37)	+0.440(32)	+0.063	+0.284	+0.348	+0.064
R (Å)	3.682(15)	3.698(15)	+0.016	3.663	3.663	+0.000
	—COOD
a (Å)	+3.008(3)	+3.018(3)	+0.010	+3.008	+3.010	+0.002
b (Å)	+1.951(4)	+1.931(4)	−0.020	+1.952	+1.944	−0.008
R (Å)	3.585(4)	3.583(4)	−0.002	3.586	3.583	−0.003

^aB3LYP/tzvp with empirical dispersion = gd3bj.

^bTD-B3LYP/tzvp with empirical dispersion = gd3bj.

In contrast, the coordinate differences are much larger between the two blue-shifted origins and differ in magnitude by a minimum of 0.8 Å and 1.5 Å, respectively. From these principal axis coordinates, the cis and trans rotamers can be readily assigned to the structures in Fig. 1. Furthermore, the coordinates indicate that the —OH groups point toward the carbonyl group which identifies the syn (s-) forms rather than the anti (a-) forms of the two conformers. Therefore, the lower energy band A may be assigned to the s-cis conformer and the higher energy band B may be assigned to the s-trans form. The assignment agrees with that of Arnold and Sulkes,² who based their assignment on the difference in lifetimes expected from the proximity effect of charge transfer quenching.^3,39 Such subtle changes in structure or the lack of them is not reliably determined from the differences in the parent’s rotational constants and, in fact, can be misleading as shown previously for the cis and trans forms of the 1- and 2-hydroxynaphthalenes.⁴⁰

IV. DISCUSSION

From the rotational analysis of I4CA and its isotopologues, we have determined the H-atom positions of the —NH and —COOH groups in the S₀ and S₁ electronic states of indole-4-carboxylic acid and have used them to unambiguously identify the conformational forms of the A and B bands as the s-cis and s-trans rotamers of I4CA, respectively. We have also determined the TDM orientations of the two forms which are shown to differ by a little as 6° or as much as 38° (the signs of the TDM angles are not determined from this analysis). The smaller line width of the s-trans conformer indicates a more than 50% increase in the fluorescence lifetime relative to the s-cis form in agreement with previously reported results.² In this section, we will now investigate these intriguing aspects of I4CA with the help of ab initio theory. These comparisons are used to identify the character of the S₁ states and give us insight into the extent of intramolecular charge transfer which impacts the fluorescence quenching dynamics of I4CA.

A. Theoretical S₀ and S₁ structures

The calculated rotational constants and S₁ origin energies at the DFT and TD-DFT levels of theory are shown together with the experimental values in Table IV. The predicted origin energy difference of the rotamers is 482 cm⁻¹ vs the observed value of 384 cm⁻¹. For both conformers, the calculated rotational constants are within 1% of the observed values in both electronic states. The agreement with the predicted S₁ ← S₀ changes is less quantitative, especially given the over- and underestimated magnitudes of ΔA and ΔB, respectively, for both conformers. Nevertheless, the overall trends are correct.

TABLE IV.

Observed and calculated rotational constants and relative energies of the s-cis and s-trans conformers of I4CA in their S₀ and S₁ electronic states. The calculated results are for planar (C_S) structures (i.e., ΔI = 0). The results for optimized C₁ structures are nearly identical.

	s-cis			s-trans
Expt.	S0	S1	S1–S0	S0	S1	S1–S0
A″/A′/Δ (MHz)	1613.2	1601.8	−11.38	1591.0	1582.7	−8.27
B″/B′/Δ (MHz)	866.6	860.4	−6.18	876.8	866.4	−10.40
C″/C′/Δ (MHz)	564.1	560.0	−4.12	565.7	560.2	−5.52
Theory	DFTa	TD-DFTb		DFTa	TD-DFTb
A″/A′/Δ (MHz)	1622.3	1595.6	−26.75	1599.7	1575.1	−24.57
B″/B′/Δ (MHz)	868.0	866.8	−1.16	878.6	874.9	−3.71
C″/C′/Δ (MHz)	565.4	561.7	−3.77	567.1	562.5	−4.65
	ΔE			ΔE		ΔEt-cc
Expt.(cm−1)	31 122			31 505		384
Theory (cm−1)d	27 100			27 582		482

^aB3LYP/tzvp with empirical dispersion = gd3bj.

^bTD-B3LYP/tzvp with empirical dispersion = gd3bj.

^cOrigin energy difference: s-trans–s-cis.

^dThe calculated DFT energy difference (s-trans–s-cis) with zero-point corrections is −94 cm⁻¹.

The calculated hydrogen atom positions of the —NH and —COOH groups in S₀ and S₁ are given beside the observed values in Table III. The coordinate signs are included for both the observed and calculated values in reference to the structures in Fig. 1. For the ground states, the predicted coordinates are nearly always within ±0.03 Å of the observed values, while the S₁ values are slightly worse but nearly always within ±0.04 Å. A few exceptions are seen for coordinates close to the b-axis where the experimental uncertainties are large. The S₁ ← S₀ differences are also included in Table III. In all cases, the observed differences are within ±0.016 Å of the predicted values. For both experiment and theory, the largest changes in S₁ occur in the b-coordinates, especially for the N—H groups. It should be noted that these changes do not necessarily reflect differences in the H-bond lengths. Additional experiments on the ¹⁵N and ¹⁸O isotopologues would be required for those determinations.

B. TDM orientations and S₁ state character

The TDM orientation of an electronic transition is a sensitive function of the frontier molecular orbitals (MOs) that define it. Rotationally resolved UV studies give a direct measure of the TDM orientation and therefore insight into the spatial character of the MOs involved. For planar molecular systems, the absolute value of the TDM angle is found directly from the overall fitted intensities, I_a and I_b, that include the line strength factors using tan(θ_TDM) = (I_b/I_a)^0.5. From the rotationally resolved study of indole,⁹ the TDM was found to make an angle of +38.3° relative to the a-axis of the parent molecule (see Fig. 4). In this case, the positive sign designating counterclockwise rotation was suggested by Philips and Levy⁴¹ from comparative studies with the TDM of tryptamine.³⁰ Upon substitution in the 4-position with the —OCH₃¹³ group, the reported S₁ TDM angle of −17° was shown to consist primarily of ¹L_b state character while the S₁ state of 4-cyano indole¹² with a TDM of ±31° was assigned to an ¹L_a state. An interesting exception was found in the study of 4-fluoroindole¹⁴ where an observed S₁ TDM angle of +63° was shown not to be representative of the ¹L_b character assigned based on the permanent dipole moment change. From the current results on I4CA, the s-cis and s-trans TDM angles from Tables I and II are primarily a-axis polarized with angles of ±15° and ±21°, respectively.

FIG. 4.

The TD-DFT optimized S₁ structures and S₁ and S₂ TDM orientations of the two conformers of I4CA and indole are shown in the top panel. The experimentally determined TDM orientations are superimposed on the DFT optimized structures in the lower panel. The signs of the experimental TDMs of I4CA are not determined in this analysis. All structures are shown in the a/b-principal axis frames.

To obtain a better picture of how these angles compare within the frame of indole, the observed TDM orientations of I4CA and indole are shown in the lower panels of Fig. 4. The a/b principal axes are also shown. Note first that because of the simple mass effect of the —COOH group (which rotates the ab-frame by >120° relative to that of indole), the TDMs of I4CA are oriented roughly parallel to the points of attachment of the acid groups. Note also that the orientations (regardless of sign) are nearly orthogonal to the TDM of indole. For comparison, the theoretical TDM angles from TD-DFT of the S₁ and S₂ states of I4CA and indole are shown in the top panels of Fig. 4. For indole, the angles relative to the a-axis are −42° and +51° for the S₁ and S₂ states, respectively. However, the observed TDM angle for the S₁ state of indole is +38.3°.⁹ Therefore, the predicted state order for indole is reversed, and the corrected labels are shown in Fig. 4. The state reversal is not uncommon and likely a result of the absence of multi-reference configuration interaction at the TD-DFT level.^10,12,42 In contrast, the close agreement with the observed orientations of I4CA indicates that the state order is correctly predicted for the two conformers of I4CA. Furthermore, it is tempting to assign positive signs for the TDMs, given the positive angles predicted for both rotamers, and the agreement with the observed angle increases for the s-trans form relative to the s-cis rotamer.

Further insight into the factors responsible for the large difference in the TDM angles of I4CA and indole is provided by the frontier molecular orbitals (MOs) shown in Fig. 5. The principal MOs involved in the S₁ and S₂ transitions are the HOMO-1, HOMO, LUMO, and LUMO+1 MOs. As shown schematically for indole on the right side of Fig. 5, the transition associated with the observed S₁ state has its principal parentage in the two one-electron excitations, LUMO ← HOMO-1 and LUMO+1 ← HOMO. Therefore, in the notation of Platt,⁸ the S₁ state has the character of the ¹L_b state as do the vast majority of the substituted indoles. In contrast, the S₁ states for both conformers of I4CA largely consist of the LUMO ← HOMO excitation that is characteristic of the ¹L_a state. A closer look at the I4CA MOs suggests some possible reasons for the state reversal. Three of the four MOs have significant amplitude on the carbonyl carbon atoms. The exception is the HOMO-1 MO where little amplitude is seen on this atom. Consequently, the enhanced MO overlap in the ¹L_a state’s LUMO ← HOMO excitation may lower its energy relative to the ¹L_b state’s LUMO ← HOMO-1 excitation. The latter excitation carries more than 65% of the component amplitudes of the S₂ transitions in both rotamers. We further note that there exists a sizable amplitude component (11%) of the LUMO ← HOMO-3 excitation (see Fig. S5) in both the S₁ and S₂ transitions of I4CA. For the S₂ state only, the component has opposite sign relative to the ¹L_b excitations shown in Fig. 5. As seen in Fig S5, these components increase the charge transfer to the carbonyl group which may further stabilize the LUMO in S₁ and/or destabilize it in S₂ and, therefore, contribute to the state reversal. In addition to the ¹L_a assignments of S₁ in 4-and 6-cyanoindole,¹² a similar ¹L_b/¹L_a state reversal was reported for 1-aminonaphthalene³⁶ relative to several other 1-substituted naphthalenes, which interestingly includes 1-naphthoic acid.⁴³ As apparently true for the latter case, the acid group interaction with the indole ring is somewhat enhanced relative to this group’s interaction with the naphthalene frame.

FIG. 5.

Frontier molecular orbitals (MOs) of the two conformers of I4CA and indole. The principal transitions associated with the ¹L_a and ¹L_b states are also shown.

C. Natural linewidths and charge transfer quenching

A fascinating aspect of the jet-cooled I4CA data is the >50% difference in the fluorescence lifetimes observed for the two rotamers.² This difference is significant in comparison with those of numerous other tryptophan analogs, in which gas-phase lifetime differences of more than 10% are uncommon. We note that these trends pertain to the origin bands only since shorter lifetimes are expected with an increase in vibrational energy in S₁ as reported for indole.¹¹ The strong conformer dependent lifetimes in I4CA are also in contrast to the two conformers of the 5-hydroxy⁴⁴ and 6-methoxy⁴⁵ indole derivatives, which display no difference in their lifetimes. Furthermore, the lifetimes of I4CA are also noticeably shorter compared to other substituted indoles with the exception of 4-hydroxyindole.⁴⁴

In the condensed phase, quenching of the fluorescence of indole derivatives by carbonyl groups has been known for some time. In a study by Ricci and Nesta,⁴⁶ and from prior results summarized there, a Stern–Volmer analysis was used to determine the fluorescence yield with and without quencher to obtain the nonradiative rate constants for charge transfer (CT) from the excited indole chromophore to various electrophilic carbonyl substituents and complexes. A trend clearly identified for different electron withdrawing groups, R and R′, of R—(C═O)—R′ was that the quenching ability is directly proportional to the electrophilicity of the carbonyl group. This relationship was especially conclusive for the acids (R′═OH), given their measurable pK_a values. Later, Petrich and co-workers¹⁵ examined the nonexponential fluorescence decay of tryptophyl compounds in solution using a time correlated single-photon counting technique where they also concluded that CT is the dominant mode for nonradiative decay. In a follow-up study,³⁹ they applied a model developed by Hopfield⁴⁷ to link the CT rate to the proximity and orientation of the donor with respect to the acceptor group, the ionization potential of the donor, and the electron affinity of the acceptor. The enhanced quenching ability of the carbonyl group is commonly rationalized within the context of oxygen’s non-bonding orbitals, n, that promote nonradiative decay through the intersystem crossing pathways, S(n, π*) → T(π, π*) and S(π, π*) → T(n, π*).⁴⁸

As these studies suggest, the short lifetimes of the two rotamers of I4CA may principally arise from CT from the indole ring to the electrophilic carbonyl group, thereby enhancing the nonradiative decay rate that competes with the radiative component.^2,3 If CT quenching is the principal nonradiative route in I4CA, then the relative magnitudes of CT may explain the large lifetime difference observed in the two rotamers. The current study performed in the cold, isolated conditions of a free jet offers unique insight into this issue without the complicating factors associated with the solution phase dynamics that alter the proximity and orientation on the donor and acceptor groups.

A semi-quantitative measure of intramolecular CT may be evaluated from the change in the atomic charges of the —COOH group upon electronic excitation. The charge analysis method chosen here is based on natural population analysis (NPA). As discussed for CT processes in molecular and metal complexes,⁴⁹ the NPA analysis seems less dependent on basis set size compared to other charge partitioning methods. Figure 6 shows the calculated natural charges on the —COOH and —NH groups in the S₀ states and the S₁ states using the DFT and TD-DFT methods. The changes relative to the ground state are given in parentheses in the upper panels of Fig. 6. The full charge sets are shown in Figs. S6 and S7 of the supplementary material for the S₀, S₁, and S₂ states.

FIG. 6.

Selected natural charges of the s-cis (left panels) and s-trans (right panels) conformers of I4CA in the S₀ (lower panels) and S₁ (upper panels) states calculated at the DFT and TD-DFT levels of theory. The S₁–S₀ changes in the charges are shown in parentheses.

For the ground states in Fig. 6, the natural charges (in atomic units, e⁻) of the two rotamers are seen to be nearly identical with differences not exceeding 0.002. While the net charges on the —COOH groups are slightly negative (−0.024 for s-cis and −0.021 for s-trans), the carbonyl atoms have net charges of +0.157 for s-cis and +0.159 for s-trans forms as expected for electrophilic groups. Consequently, upon excitation, significant charge transfer occurs to the —COOH groups of both rotamers, giving net charges of −0.218 and −0.209 and changes relative to the ground state of −0.194 vs −0.188 for the s-cis and s-trans rotamers, respectively. Most relevant to the >50% lifetime difference is the change in the carbonyl groups’ net charges relative to S₀. For the s-cis rotamer, the CT difference is −0.172, which represents a 25% increase relative to −0.135 for the s-trans form. As argued above, this larger CT for the s-cis rotamer is expected to enhance the nonradiative decay rate and may, in part, explain its shorter lifetime.

Charge transfer to the acid groups is also reflected in the changes in the S₁ bond lengths, some of which are shown in Fig. 7. For example, the increase in the negative charge on the carbonyl groups increases the bond lengths in S₁ by +0.024 Å for the s-cis form compared to +0.017 Å for the s-trans form. Similar increases are seen for the C—OH bond but reversed in magnitude for the two rotamers: +0.016 Å for s-cis vs +0.032 Å for the s-trans forms. Moreover, the increase in electron density in S₁ will make these groups stronger bases which may lead, in turn, to the predicted decreases in the hydrogen bonding distances between the oxygen atom of the acid groups and the adjacent H atom of the pyrrole ring, −0.103 Å for s-cis and −0.084 Å for s-trans. The increase in the H-bond strength should reduce the out-of-plane vibrational contributions of the acid group and may partly explain the observed increases in planarity in the S₁ states (i.e., the S₁ inertial defects are less negative relative to those in S₀, see Tables I and II).¹¹ These results also support the rotamer assignment of Arnold and co-workers⁴ that was based on the expected fluorescence lifetime dependence on the proximity of the carbonyl and pyrrole ring moieties.⁵ Finally, we note that similar decreases in the charge density on the N—H groups (see Fig. 6) for both rotamers in S₁ lead to increases in the bond lengths of 0.005 Å, suggesting a slight increase in the N—H groups’ acidity relative to S₀. These N—H bond length increases in S₁ are more than 3-fold smaller than the observed H-atom center-of-mass increases of ≈+0.016 Å from Table III.

FIG. 7.

Selected bond lengths in Å of the s-cis (left panels) and s-trans (right panels) conformers of I4CA in the S₀ (lower panels) and S₁ (upper panels) states calculated at the DFT and TD-DFT levels of theory. The S₁–S₀ changes in the bond lengths are shown in parentheses.

V. CONCLUSIONS

High resolution S₁ ← S₀ fluorescence excitation experiments on the $0_0^0$ bands of the s-cis and s-trans conformers of I4CA and their —ND and —COOD isotopologues have been performed in the collision-free environment of a molecular beam. Rotational constants obtained from fits have been used to determine the substituted atom coordinates in both electronic states and to make definitive identification of the s-cis and s-trans conformational forms. The observed TDM orientations relative to the indole frame are found to be nearly orthogonal to that of the S₁(¹L_b) ← S₀ origin transition in indole. Comparing these results with the predicted TDM orientations from TD-DFT indicates that the S₁ states of I4CA are primarily ¹L_a in character (LUMO ← HOMO) with some additional component, LUMO ← HOMO-3, contributing to intramolecular CT. The observed state reversal relative to other substituted indoles is unusual and suggests the importance of CT to the acid group as a stabilizing factor for the ¹L_a state. Further evidence for this interpretation comes from the changes in the natural charges and bond lengths upon photoexcitation. From the calculated charges on the carbonyl groups in S₀ and S₁, the propensity for CT in S₁ is enhanced by ≈25% in the s-cis rotamer relative to the s-trans form. The larger CT in the former is expected to increase its nonradiative quenching rate and, hence, shorten its lifetime relative to the s-trans rotamer, as experimentally observed. To further elucidate CT effects in the rotamers of I4CA, on-going studies of the NH₃ and H₂O complexes are expected to provide detailed information about the changes in acidity of the —COOH groups upon electronic excitation and the absolute orientation of the TDM through quantum state interference effects.⁵⁰ Other studies may include the ¹⁸O enriched samples of the acid groups to provide for more rigorous comparisons with the theoretical S₁ bond length changes.