Femtosecond Spectroscopy of Calcium ... - ACS Publications

Aug 29, 2016 - 1, D-40225 Düsseldorf, Germany. ‡. Institut für Theoretische Chemie und Computerchemie, Heinrich-Heine-Universität Düsseldorf, ...
0 downloads 0 Views 3MB Size
Article pubs.acs.org/JPCB

Femtosecond Spectroscopy of Calcium DipicolinateA Major Component of Bacterial Spores Ramona Mundt,†,§ Christian Torres Ziegenbein,†,§ Sascha Fröbel,† Oliver Weingart,‡ and Peter Gilch*,† †

Institut für Physikalische Chemie, Heinrich-Heine-Universität Düsseldorf, Universitätstr. 1, D-40225 Düsseldorf, Germany Institut für Theoretische Chemie und Computerchemie, Heinrich-Heine-Universität Düsseldorf, Universitätstr. 1, D-40225 Düsseldorf, Germany



S Supporting Information *

ABSTRACT: Bacterial spores are rich in calcium dipicolinate (CaDPA). The role of this compound in the high UV resistance of spore DNA and their unique DNA photochemistry is not yet clarified. Here, the photophysical properties of CaDPA dissolved in water are studied by means of steady-state and time-resolved spectroscopy as well as quantum chemistry. Upon 255 nm excitation, a fluorescence emission with a yield of 1.7 × 10−5 is detected. This low yield is in line with a measured fluorescence lifetime of 110 fs. Transient absorption experiments point to further transitions with time constants of 92 ps and 6.8 μs. The microsecond time constant is assigned to the decay of a triplet state. The yield of this state is close to unity. With the aid of quantum chemistry (TD-DFT, DFT-MRCI), the following transitions are identified. The primarily excited 1ππ* state depletes within 110 fs. The depletion results in the population of an energetically close lying 1nπ* state. An El-Sayed allowed intersystem crossing process with a time constant of 92 ps ensues. Implications of these findings on the interaction between photoexcited CaDPA and spore DNA are discussed.

1. INTRODUCTION Certain bacteria, with bacillus and clostridium as the two major species,1,2 may transform into (endo)spores under environmental stress.3 These dormant structures of bacteria exhibit high resistances against wet and dry heat, desiccation, UV- and γ-radiation, as well as chemical agents.4 These resistances and lifespans exceeding centuries5−7 pose serious health and security risksthe pathogenic agent of anthrax is a spore forming bacterium.8 The photobiology of DNA within spores differs from DNA in the vegetative form. The spore DNA exhibits a higher UV resistance,9−11 and the predominant UV lesion is the spore photoproduct.10,12 In the vegetative form and in other organisms, the predominant lesions are cyclobutane pyrimidine dimers (CPD) and (6−4) lesions.12 Despite 50 years of research,11,12 the molecular mechanisms underlying the formation of the structurally crossly different spore photoproducts are not well understood. There seems to be consensus that three factors contribute to this behavior.11 The core of spores, which houses the DNA, is poor in water content and rich in small, acid-soluble spore proteins (SASPs) and the calcium salt of dipicolinic acid (CaDPA; for structure, see Figure 1). The two former factors seem to favor a predominance of A-like DNA.13 In contrast, in the vegetative form, B-DNA prevails.11 CaDPA, making up to 20% of the dry weight of the spore core,11 could also affect the ground state conformation of the © 2016 American Chemical Society

DNA and/or interact with the DNA via its excited states. The absorption spectrum of CaDPA14 (see Figure 1) overlaps with that of the DNA bases (see, e.g., ref 15) and, thus, might potentially act as a “sunscreen”. Seemingly in line with this, CaDPA-less spores were found to be more UV-sensitive than CaDPA-rich ones.16 On the other hand, evidence was given that excited CaDPA can donate electronic energypresumably via triplet excitationsto thymine bases,17−19 which would have an adverse effect on spore DNA photostability. The observation of an energy transfer renders it at least plausible that excited states of CaDPA play a role in the unique photobiology of spore DNA. Desnous et al.12 state in their review that “full clarification would first require a detailed examination of the excited-state properties of CaDPA”. This contribution is considered a first step toward such a “full clarification”. With relation to noncontact detection of anthrax, many spectroscopic studies on the ground state of CaDPA were published.20,21 However, concerning the excited states, not even a (correct) fluorescence spectrum was reported. This can be attributed to the small fluorescence quantum yield of the compound22 and, more importantly, to a photoreaction resulting in a product with a highly increased yield.22 Received: June 20, 2016 Revised: August 11, 2016 Published: August 29, 2016 9376

DOI: 10.1021/acs.jpcb.6b06230 J. Phys. Chem. B 2016, 120, 9376−9386

Article

The Journal of Physical Chemistry B

Figure 1. UV/vis absorption (absorption coefficient ε versus wavelength λ) in H2O (pH 7.5) and fluorescence spectra (∝ photons per wavelength interval) of CaDPA (structure on the right) in D2O (pD 8). To record the fluorescence spectrum, the excitation was tuned to 255 nm. The spectral resolution of the fluorescence experiment was ∼5 nm.

The protocol adopted here is detailed in the Materials and Methods section. 2.2. Steady-State Spectroscopy. For the absorption measurements presented here, H2O was used as solvent. CaDPA exhibits an absorption band lowest in transition energy peaking around 270 nm (Figure 1). In line with earlier findings,28 we observe a sharper vibronic progression for Ca2+ as a counterion in comparison to Na+ ions (data not shown). In CaDPA, the spacing of the progression amounts to 1081 cm−1. The sharpening may be attributed to the formation of a rather rigid chelate complex.28 The transition is of moderate strength. The absorption coefficient ε270nm at the maximum amounts to 5170 M−1 cm−1. The respective experimental oscillator strength f equals 0.09. For the computation of this value (see ref 29), the upper and lower bounds of the integration were set to 355 and 245 nm, respectively. The fluorescence spectra were recorded in D2O, and the excitation wavelength was tuned to 255 nm. For these measurements, the combination of deuterated solvent and low excitation wavelength prevents the solvent Raman peak from overlapping with the CaDPA fluorescence signature. Additional fluorescence experiments with H2O as a solvent and with varying excitation wavelengths did not reveal significant differences in the spectroscopic properties. The CaDPA fluorescence spectrum peaks at 292 nm, and the Stokes shift, thus, amounts to 2790 cm−1. The spectrum does not exhibit any vibronic structure. In the spectral range 320−330 nm, the short wavelength tail of a second band is seen. This band is attributed to the photoproduct. Its amplitude rises with decreasing flow rate of the sample solution, since at lower flow rates the concentration of photoproducts in the focal volume increases. To determine the fluorescence quantum yield ϕfl of CaDPA, its fluorescence signal was measured in comparison with that of thymidine in water. The one for CaDPA is by a factor of 5.8 smaller than the one for thymidine. Relying on the fluorescence quantum yield ϕfl,r of 10−4 for thymidine,30,31 one obtains a minute value of 1.7 × 10−5 for the yield ϕfl. On the basis of this yield and the radiative rate constant kr, which can be obtained via a Strickler−Berg analysis,29,32 the fluorescence lifetime τfl,sb may be estimated using

We here report femtosecond (fs) and nanosecond (ns) UV/ vis transient spectroscopy on CaDPA. By fs fluorescence spectroscopy, the ultrafast decay of the primary excitation will be traced. Further transiently populated excited states are identified by fs and ns absorption spectroscopy. Assignments of the spectroscopic signatures will be backed by quantum chemistry. Finally, implications of the results on the DNA− CaDPA interaction will be discussed.

2. RESULTS 2.1. Defining the Sample. Dipicolinic acid (DPA) contains two acidic moieties and one basic moiety. Thus, depending on the pH, different protonation forms exist in solution. For pH values larger than ∼5.2 (pKa,3), the dianionic form DPA2− prevails.23 The pH within the core of spores was reported to be around 6.4,24 and the dianionic form should, thus, dominate. For the experiments presented here, solutions with slightly basic pH or pD values (7−8) were used. This ensures that the concentration of the dianionic form is larger than those of the other ones by at least 2 orders of magnitude. We did not use buffers or solutions with even higher pH values, since anionic species facilitate the generation of hydrated electrons.25 The spectroscopic signatures of this generation can superimpose onto the “genuine” signal. Spectroscopic studies26,27 suggest that in the spores the stoichiometry of Ca2+ and DPA2− is 1:1. Thus, in the experiments, focus will be laid on solutions with a 1:1 stoichiometry. The spectroscopic impact of the counterions is studied by comparing solutions of CaDPA with ones of Na2DPA. The concentration of CaDPA in the spore core (∼1 M) exceeds its solubility in water by an order of magnitude.11 Thus, CaDPA in the core might be a crystalline or glass-like solid. In this first study on the excited state properties of CaDPA, we will focus on the molecular behavior in solution. Such studies (and those on solids even more) are hampered by a photoreaction of the compound.22 In the presence of air, we determined a quantum yield ϕr for this reaction of ∼2 × 10−3. Purging with nitrogen reduces this yield to ∼5 × 10−4. Both values refer to the consumption of the starting material. The photoproduct, which was not further characterized, exhibits a fluorescence quantum yield ϕfl of 0.025, which is much higher than the one of DPA (see below). Thus, without a constant exchange of the sample solution using a flow system, steady state and time-resolved fluorescence data cannot be obtained.

τfl,sb = 9377

ϕfl kr

(1) DOI: 10.1021/acs.jpcb.6b06230 J. Phys. Chem. B 2016, 120, 9376−9386

Article

The Journal of Physical Chemistry B

Figure 2. fs transient fluorescence on CaDPA in D2O (∼2 mM) as a function of detection wavelength λ and delay time td. The solution was excited at 255 nm. The spectral resolution was ∼5 nm. In the central contour representation, purple hue represents large fluorescence signals. One representative time trace (300 nm) and the IRF are plotted on the left. Experimental data as well as fit results (lines) are given. One selected spectrum (delay time td = 100 fs) is depicted on the right.

Figure 3. fs transient absorption on CaDPA as a function of detection wavelength λ and delay time td. The solution was excited at 266 nm. The pH was around 7.5, and concentrations were around 6 mM. In the central contour representation, reddish hue stands for positive difference absorption due to ESA and bluish coloring represents negative contributions due to GSB and SE. The time axis is linear until 1 ps and logarithmic thereafter. Representative time traces are plotted on the left (vertical dashed lines in the contour plot indicate their spectral position). Selected difference spectra are depicted on the right. Their vertical positions (horizontal dashed lines) correspond to the respective delay time.

Figure 4. fs transient absorption on Na2DPA as a function of detection wavelength λ and delay time td. The solution was excited at 266 nm. The pH was around 7.5, and concentrations were around 6 mM. In the central contour representation, reddish hue stands for positive difference absorption due to ESA and bluish coloring represents negative contributions due to GSB and SE. The time axis is linear until 1 ps and logarithmic thereafter. Representative time traces are plotted on the left (vertical dashed lines in the contour plot indicate their spectral position). Selected difference spectra are depicted on the right. Their vertical positions (horizontal dashed lines) correspond to the respective delay time.

solutions were excited with fs pulses centered at 255 nm. Fluorescence emission as a function of the detection wavelength λ and the delay time td was probed by Kerr gating. With the Raman scattering of neat D2O as a signal, the instrumental response function (IRF) was recorded. Its temporal width (fwhm) equals 270 fs. The time-resolved spectra match the

The Strickler−Berg analysis using the spectra plotted in Figure 1 as input yields a value of 9.8 × 107 s−1 for the rate constant kr. This implies an estimated fluorescence lifetime τfl,sb of 170 fs. 2.3. fs Fluorescence Spectroscopy. The lifetime τfl derived from steady-state spectroscopy is confirmed by timeresolved fluorescence spectroscopy (Figure 2). Sample 9378

DOI: 10.1021/acs.jpcb.6b06230 J. Phys. Chem. B 2016, 120, 9376−9386

Article

The Journal of Physical Chemistry B shape of the steady state one, except for the second band mentioned above. Its absence underscores the above notion that this second band is due to fluorescence of the photoproduct. No spectral changes due to vibrational or dielectric relaxation are discernible. The fluorescence lifetime τfl was determined by a global fit of the data employing a single exponential trial function convoluted with the IRF (see Materials and Methods). The value of 108 fs ± 10% matches the above estimate. We stress that the value is well-defined despite the fact that it is of the order of the IRF time. The fluorescence lifetime τfl has two effects on the recorded time traces (cf. Figure 2, left). It broadens them in comparison to the IRF and shifts the maximum to positive delay times. The fitting routine “uses” both effects to pin down the lifetime. 2.4. fs Absorption Spectroscopy. States populated after the fluorescence decay are identified by fs absorption spectroscopy (Figure 3). The CaDPA solution was excited with 266 nm laser pulses, and the resulting absorption changes ΔA were probed by a white light continuum. Around time zero, a strong band at the high energy edge of our detection window (320 nm), a weaker one centered at 440 nm, and a weak and very broad one at 700 nm are discernible. Up to ∼100 ps, only slight spectral changes are observable. In particular, the ∼100 fs process observed in the fluorescence experiment does not cause a strong imprint in the fs absorption data. The three bands decay simultaneously on the 100 ps time scale giving way to an offset spectrum. The latter is characterized by a shoulder at 380 nm and a very flat positive difference absorption throughout the visible range. A very similar spectro-temporal pattern is observed for solutions of Na2DPA (Figure 4). In a study on lanthanoide complexes of DPA2− in acetonitrile,33 except for the 100 fs component, similar transient spectra and comparable lifetimes have been observed. To determine the time constants, the data set was subject to a global analysis using a multiexponential trial function convoluted with the IRF (see Materials and Methods). From the inspection of the data (Figure 3), it becomes obvious that one term is needed to describe the pronounced decay on the 100 ps time scale. The time traces further suggest a term accounting for processes on the 1−10 ps time scale. Finally, the fluorescence data has given very clear evidence for an ∼100 fs process. We, thus, opted for a trial function with three exponentials and an offset. With this minimum number of exponential functions, a satisfying fit was possible. The time +45% constants retrieved are τ1 = 110+55% −40% fs, τ2 = 1.1−30% ps, and τ3 = +10% 92−10% ps. The τ1 value is in good agreement with fluorescence data. The respective decay associated spectra (DAS) ΔA1,2,3,4(λ) are compiled in Figure 5. The sum of these spectra (cf. Figure 5) represents the time zero spectrum corrected for the finite time resolution (cf. eq 4). The DAS ΔA1(λ) features negative bands at around 320 and 440 nm. Since in this range neither ground state bleach (GSB) nor stimulated emission (SE) can contribute, these bands parametrize rises of excited state absorption (ESA). These rises are difficult to distinguish visually from the rise caused by the instrumental response, which explains why no imprint of the τ1 process is visible in Figure 3. The DAS ΔA2(λ) has a sigmoidal pattern with a zero crossing at 450 nm. There is a hint for a second pattern with a crossing at 340 nm. Sigmoidal patterns may represent spectral shifts due to relaxation processes.34,35 The pattern here represents a blue shift of bands centered around the zero crossings. The DAS ΔA3(λ) parametrizes the pronounced decay on the 100 ps time scale described above. The DAS

Figure 5. DAS retrieved from a global analysis of the CaDPA data set depicted in Figure 3. The uppermost spectrum is the sum of all DAS. Starting from here, the color coding in all figures is based on the assignments in Figure 10; that is, blue is used for spectra related to the S3 state, green to the S1 and S2 states, and red to the T1 state.

ΔA4(λ), finally, is identical to the spectrum recorded 1 ns after photoexcitation (cf. Figure 3). The analysis of the Na2DPA data yields time constants very close to those obtained for CaDPA. The only marked difference is a reduced (relative) amplitude of the offset signal. It amounts to only one-half of the CaDPA value. 2.5. ns Absorption Spectroscopy. To identify the carrier of the offset spectrum, ns absorption data were recorded. 7 ns laser pulses centered at 266 nm excited the sample solution, and the absorption changes at various wavelengths were probed by a pulsed xenon lamp. Single exponential decays were observed. By a global analysis, the DAS ΔAns(λ) was obtained (Figure 6). One exponential decay term sufficed to properly fit the data. In the range 320−400 nm, it overlays with the DAS ΔA4(λ) obtained from fs spectroscopy (cf. Figure 5). Beyond the spectral range accessible by our fs instrument, the DAS features a maximum at 290 nm and a minimum at 270 nm. The minimum is caused by the ground state bleach superimposed onto a positive difference absorption. The lifetime of the transient depends on the oxygen content of the solvent. For water purged with nitrous oxide, the lifetime τns amounts to 6.8 μs. In air-saturated water (oxygen concentration [O2] of 0.29 mM),36 the lifetime is reduced to 1.3 μs. Purging with neat oxygen at 1 atm (1.39 mM)36 further reduces this lifetime to 0.3 μs. Analyzing this behavior according to 9379

DOI: 10.1021/acs.jpcb.6b06230 J. Phys. Chem. B 2016, 120, 9376−9386

Article

The Journal of Physical Chemistry B

out the importance of singlet nπ* states for the depletion of the primary excitation and intersystem crossing (ISC) processes yielding triplet states. Treatment of the excitations in CaDPA poses some additional challenges. While in pyridine only the lone pair at the nitrogen atom requires considerations when it comes to the nπ* excitation, in CaDPA, lone pairs at the carboxylate substituents and the nitrogen atom may contribute. For a dianion like DPA2−, one may expect more diffuse molecular orbitals (MOs) as compared to a neutral molecule and this may require larger basis sets.40 Finally, the exact structure of the chelate complex CaDPA in solution is not known. Here, as a starting point, the DPA geometry of the crystal structure of CaDPA11,41 was used. In the crystal, CaDPA exists as a dimer. The carboxylate oxygens contribute to the ligand sphere of two Ca2+ ions. Water ligands also bridge the ions. In solution, no indications for an association of DPA2−, i.e., without Ca2+, were found at high (basic) pH values.23 In the computations, a monomer was treated (see Figure 7). In this monomer, the Ca2+ ion is coordinated by a planar DPA2− moiety. Two carboxylate oxygens as well as the nitrogen of the pyridine coordinate the central ion.

Figure 6. Amplitude spectrum ΔAns(λ) derived from ns absorption spectroscopy (red solid line). The transient spectrum of the fs experiment at 3 ns is overlaid (red dashed line). The inverted ground state spectrum (black line) is trimmed into the ns data to obtain the absorption coefficients. The triplet spectrum corrected for the ground state bleach is also given (red dotted line).

1 = k 0 + kq[O2 ] τns

(2)

yields a unimolecular decay constant k0 of 1.7 × 105 s−1 and a quenching constant kq of 2 × 109 M−1 s−1. Both values are in the range characteristic for organic triplet states.36 The quantum yield of this triplet state ϕt is obtained as follows. In the first step, difference absorption coefficients Δε(λ) were computed from DAS ΔAns(λ) (cf. Figure 6). To this end, the absorption spectrum ε(λ) of CaDPA (cf. Figure 1) was redrawn to match the spectral resolution and data spacing of the ns instrument. This spectrum was then inverted, shifted, and scaled so that it fits into the minimum of the DAS ΔAns(λ) (cf. Figure 6, black line). The peak absorption coefficient of the redrawn spectrum amounts to 4808 M−1 cm−1 (symbolized by the gray arrow in Figure 6). With this coefficient as a ruler, a Δε axis can be drawn. By subtraction of the ground state bleach, the triplet spectrum is obtained which peaks at 280 nm. In the second step, signals were recorded for naphthalene dissolved in cyclohexane37 and CaDPA. Care was administered to ensure identical excitation conditions (laser power and absorption at the excitation wavelength). For these conditions, the yield ϕt can be computed via Δεr,414nm ΔA ϕt = ϕt,r 300nm Δε300nm ΔA r,414nm

Figure 7. BH-LYP orbitals of the CaDPA complex contributing to excitations in Table 1.

Geometry optimizations of the neutral single monomer CaDPA complex have been performed in singlet and triplet states with the Gaussian 09 program.42 To treat solvent effects, the polarizable continuum model (PCM) was applied.43 The quality of the calculations strongly depends on the chosen method for ground and excited state geometry optimizations. Mennucci and co-workers point out that reliable excited state structures should be computed at the CAM-B3LYP level of theory rather than B3LYP to include long-range interactions with the solvent.44 After evaluation for the system at hand, this functional was used in all geometry optimizations in combination with the 6-311++G** basis set. Excited state geometries were obtained with the time-dependent (TD) DFT formalism. A comparison of various functionals and essential structure parameters can be found in the Supporting Information (section 1.1). Spectral properties of ground and excited state species were computed with the DFT-MRCI extension 45,46 to the

(3)

According to ref 37, the triplet yield ϕt,r of the reference naphthalene equals 0.75. Its difference absorption coefficient Δεr,414nm at the peak amounts to 24 500 M−1 cm−1. The difference absorption coefficient Δε300nm of CaDPA is given in Figure 6 (4861 M−1 cm−1). Inserting the measured values of the difference absorption ΔA300nm and ΔAr,414nmtypical numbers were 7.5 and 35.4 mOD, respectivelyone arrives at a triplet yield ϕt of 0.8 ± 0.2. 2.6. Quantum-Chemical Computations. A quantum chemical treatment of CaDPA may use earlier studies on its parent compound pyridine as guidance.38,39 The dianion DPA2− has been treated by TD-DFT methods in an earlier work by Xie et al. in vacuo, reporting good agreement with experimental spectral data.40 The studies on pyridine38,39 point 9380

DOI: 10.1021/acs.jpcb.6b06230 J. Phys. Chem. B 2016, 120, 9376−9386

Article

The Journal of Physical Chemistry B

Table 1. Ground State Energy and Optical Properties of the First Three Excited Singlet States and First Six Triplet States in CaDPAa state S0 S1 S2 S3

T1 T2 T3 T4 T5 T6

excitation GS (−1301.920946 H) H-2 → H-1 → H-1 → H-2 → H → H-5 → H-3 → H → H-2 → H-3 → H-3 → H-2 → H-1 → H-1 → H-2 → H-3 → H-4 →

weight L L+1 L L+1 L L+1 L+1 L+1 L L L L L+1 L L+1 L+1 L

0.99 0.50 0.34 0.47 0.35 0.66 0.13 0.11 0.81 0.49 0.43 0.37 0.53 0.29 0.46 0.35 0.50 0.28

vert. (eV)

(cm−1)

f

type

4.76 (4.45)

38387

0.003

4.79 (4.52)

38595

0.000

4.86 (4.69)

39164

0.100

4.03 (3.95) 4.19 4.48

32504 33795 36134

4.52

36457

4.59

37021

5.01

40409

n → π* n → π* n → π* n → π* π → π* π → π* π → π* π → π* π → π* π → π* π → π* n → π* n → π* n → π* n → π* π → π* π → π*

a Orbitals (cf. Figure 7) dominantly contributing to the transitions as well as the weights (the squared coefficients) are given. All vertical excitation energies are given relative to the S0 ground state. The values in brackets denote adiabatic excitation energies. f represents the oscillator strength.

Turbomole program,47,48 applying the newly developed tight parameter set by Lyskov et al.49 The optimized ground state geometry has C2v symmetry. A normal-mode analysis reveals only positive Hessian eigenvalues; i.e., this structure corresponds to an equilibrium structure. In comparison to the computation of Xie et al.,40 which reports values for the dianion in a vacuum, we find shorter C−C, N−C, and C−O bond distances, except for the Ca2+ coordinated C− O bond which gets longer by ca. 0.02 Å (see the Supporting Information (section 1.2) for geometry data). All orbitals relevant here are centered on the dipicolinate moiety (Figure 7). Orbitals centered at the Ca2+ ion do not come into play. This may explain the similarity in the photophysics of CaDPA and Na2DPA (cf. Figures 3 and 4). According to the DFTMRCI computation, the bright excitation is due to the S0 → S3 transition which is of ππ* character and involves the HOMO and LUMO orbitals (see Table 1 and MOs in Figure 7). The computation places two singlet nπ* (S1,2) states slightly below the bright S3 state. This is true also for the adiabatic excitation energies, which were computed by optimizing the geometries of the corresponding states (italic values in Table 1). The splitting is slightly larger in the latter. The computation identifies the S3 state as the lowest excited singlet state with a noticeable oscillator strength f (0.100). This strength matches the experimental value (0.09). The computed vertical excitation energy of the S3 state is 39164 cm−1. If one equates the maximum of the absorption spectrum located at 270 nm or 37037 cm−1 (Figure 1) with the vertical excitation energy, the computation is off by 2127 cm−1. Reported corrections for vibrational zero-point motions computed for pyridine are of similar magnitude39 but do not consider solvent effects. To include zero-point corrections and temperature effects in our computations, 200 distorted structures were generated from the calculated normal modes, considering a temperature of 300 K with the local mode sampling routine provided by Gaussian 09.42 Through the geometry deformations introduced in the sampling protocol, the state energies

and oscillator strengths are slightly altered, potentially enabling access to higher lying excited states. We stress that this procedure cannot reproduce the vibronic progression observed in the experimental spectrum. The resulting spectrum, generated by convolution of individual excitation energies and oscillator strengths with a Gaussian function (fwhm = 5 nm), is shown in Figure 8. The calculated

Figure 8. DFT-MRCI computed effective UV spectrum of CaDPA in water (black), including zero-point corrections and temperature effects (see text). The vertical excitation to the S3 state is shown in blue. The oscillator strength axis on the left applies to this vertical excitation only. The experimental absorption spectrum in water (cf. Figure 1) and the corresponding absorption coefficients are added in purple.

vertical excitation is also shown. The computed spectrum shows two high intensity bands. The low energy band is due to the transition to the S3 state. Its maximum is red-shifted from the vertical excitation by ca. 7 nm (1000 cm−1). The progression aside a good agreement with the experimental spectrum is observed. The second band with higher intensity at ca. 230 nm is attributed to another singlet ππ* state (occurring as S7, S8, or S9 state) involving transitions from lower lying π orbitals (HOMO−3, HOMO−4, HOMO−5) to the LUMO 9381

DOI: 10.1021/acs.jpcb.6b06230 J. Phys. Chem. B 2016, 120, 9376−9386

Article

The Journal of Physical Chemistry B

band in the CaDPA spectrum in water (cf. Figure 1) to the S0 → S3 transition.

and LUMO+1 orbitals. A transition with a compatible oscillator strength does not occur among the computed states for the vertical spectrum. Presumably, a higher excited state is lowered in energy due to thermally accessible geometry distortions. Geometry optimization of the bright S3 state retains C2v symmetry, with all C−C and N−C aromatic bonds slightly elongated by ca. 0.02−0.03 Å. The vertical emission energy is computed using this structure. The value of 36049 cm−1 compares with an experimental value of 34247 cm−1. The difference of the computed vertical excitation and emission energies yields a Stokes shift of 3115 cm−1. The experimental value is somewhat smaller than this (2790 cm−1). This might be related to the fact that the computation relates to a completely relaxed S3 state. The experiment suggests that prior to such a complete relaxation the S3 state is depleted. For such a condition, one expects a smaller Stokes shift. To monitor how the state energies change during relaxation toward the bright state minimum, a linear interpolation from the Franck− Condon point (S0 ground state geometry) to the S3 state structure was performed in internal coordinates with DFTMRCI. The resulting profiles, showing the diabatic states (i.e., following their state character), are depicted in Figure 9. During

Figure 10. Scheme of the photophysics of CaDPA in water. Computed adiabatic energies are given in cm−1. The time constants refer to experimental ones. The molecular orbitals and the bent arrows on the left illustrate the character of the excited states.

This bright 1ππ* state is very short-lived (τ1 = 110 fs). Efficient depletion of this state via rapid internal conversion (IC) to the ground state may be excluded. Otherwise, the triplet yield ϕt could not be so close to unity. Intersystem crossing processes to five triplet states as well as IC processes to the S1 and S2 states are feasible on energetic grounds. With reference to computed spin−orbit coupling constants for pyridine,39 direct ISC from the S3 state can be considered very unlikely. The coupling constants are of the order of 10 cm−1.39 Even for most favorable (thermally averaged) Franck− Condon factors, such couplings predict lower limits for ISC time constants of 1−10 ps, which is much larger than τ1. For this prediction, the classical limit for Franck−Condon factors (Marcus type expression) was used.50 This points toward a fast IC transition to the S1 and S2 states with nπ* character. The energy profiles in Figure 9 suggest a barrier-less path from the Franck−Condon point of the S3 state to both states. For lack of vibronic coupling elements, we presently cannot clarify which of the two states is populated. The 1nπ* excitation is depleted within ∼100 ps. The nanosecond experiment gives convincing evidence that this depletion results in the population of the lowest triplet state T1. The computations assign a 3ππ* character to this state. This ISC process is, thus, El-Sayed allowed. Again, relying on the spin−orbit coupling for pyridine,39 the measured time constant is well within the expectation for such an El-Sayed allowed process. What are the implications of the presented results on the interaction of photoexcited CaDPA and DNA? As indicated in the Introduction, the spectral overlap of the absorption spectra of CaDPA and DNA has led to the suggestion that CaDPA could act as a “sunscreen”.16 Common sunscreens absorb UV radiation and rapidly transform the electronic excitation energy into heat.51 On the basis of the peak absorption coefficient of CaDPA (∼5000 M−1 cm−1, see Figure 1), its concentration in the spore core (∼1 M), and the typical diameters of such cores (