Article pubs.acs.org/JPCB
Methylene Blue Exciton States Steer Nonradiative Relaxation: Ultrafast Spectroscopy of Methylene Blue Dimer Jacob C. Dean, Daniel G. Oblinsky, Shahnawaz Rafiq, and Gregory D. Scholes* Department of Chemistry, Princeton University, Princeton, New Jersey 08544, United States S Supporting Information *
ABSTRACT: The photochemistry and aggregation properties of methylene blue (MB) lead to its popular use in photodynamic therapy. The facile formation of strongly coupled “face-to-face” H-aggregates in concentrated aqueous solution, however, significantly changes its spectroscopic properties and photophysics. The photoinitiated dynamics of the simplest MB aggregate, MB2, was investigated over femtosecond to nanosecond time scales revealing sequential internal conversion events that fully relax the excited population. MB monomer dynamics were analyzed in tandem for a direct comparison. First, ultrafast internal conversion from the electric-dipole allowed upper exciton state to the lower forbidden exciton state was evaluated by use of broadband transient absorption (BBTA) and two-dimensional electronic spectroscopy (2DES) with a time resolution of ∼10 fs. Lineshape analysis of MB and MB2 2DES bands at 298 and 77 K show effectively no difference in the diagonal/antidiagonal line width ratio for the dimer, in marked contrast to the distinct reduction of the homogeneous line width for MB. This result is interpreted as ultrafast population relaxation imposing a limitation to the homogeneous line width, instead of pure dephasing as in the case of the monomer. Narrowband transient absorption was performed with the aid of target analysis, to model the dynamics at longer times. The MB dynamics were described by a sequential model featuring vibrational relaxation (1−10 ps) followed by intersystem crossing and internal conversion (τ ∼ 370 ps) leaving behind MB triplet species. Alternatively, the dimer dynamics were entirely quenched within ∼10 ps, yielding a ground state recovery time of 3−4 ps. Such fast and complete relaxation to the ground state demonstrates the effect of concentration quenching when monomers are brought into close proximity. The formation of exciton states introduces an initial energy funnel that eventually leads to population relaxation to the ground state, preventing even the dissociation of dimers despite having internal energies well above its binding energy.
1. INTRODUCTION
While MB is known to damage a range of biological targets (DNA, proteins, lipids, and mitochondria), the exact PDT mechanism responsible for inactivation in each of those cases has been debated given the diversity of biological environments and binding motifs in vivo. Generally two mechanisms have been established: Type I − direct oxidation of biological targets by MB radicals formed following a reduction step; Type II − triplet−triplet annihilation between 3MB* (after intersystem crossing) and ground state oxygen forming singlet oxygen directly. Indeed, it has been found that MB forms singlet oxygen with a quantum yield of ΦΔ = 0.5, implying nearly complete conversion of triplets.8,12 It is therefore generally accepted that the prevalent Type II mechanism proceeds predominantly through MB monomers; however, it has been suggested that aggregates of MB themselves (MBn) can undergo electron transfer following electronic excitation ([MB2]2+ →MB• + MB2+) to form radicals in solution, thereby
Methylene blue (MB) is a cationic dye known for its range of biomedical applications, much of which is driven by its ability to aggregate with itself in aqueous media as well as other biological substrates including DNA/polynucleotides,1−3 and negatively charged interfaces (such as biological membranes).4−6 Further, its spectroscopic/photophysical properties promote its use for photosensitization in vivo since it has a large electronic absorption band peaking at 664 nm well within the “therapeutic window” of the skin. After absorption, MB efficiently undergoes intersystem crossing, with a quantum yield of ΦISC = 0.52, to form triplet species which can further sensitize singlet oxygen.7,8 These aspects of MB’s photophysics are advantageous for applications in photodynamic therapy (PDT), where the model photosensitizer is one that can be localized in a target area and subsequently activated by light to damage/destroy the target that it is bound to (or in close proximity to). In fact, MB has been shown to be effective in treating tumors and cancerous tissue such as melanoma,8 as well as viruses, fungi, and bacteria.8−11 © XXXX American Chemical Society
Received: December 3, 2015 Revised: January 9, 2016
A
DOI: 10.1021/acs.jpcb.5b11847 J. Phys. Chem. B XXXX, XXX, XXX−XXX
Article
The Journal of Physical Chemistry B promoting additional Type I reactivity.6,8,13 This latter mechanism has largely been interpreted through microsecond transient absorption studies;5,6,13 however, picosecond transient absorption experiments of MB−nucleotide complexes have revealed that those aggregates return to their ground state in fewer than 10 ps.1,2 Furthermore, additional reports have peripherally suggested that the MB2 dimer, which is efficiently formed with increasing concentration in aqueous solution, undergoes nonradiative relaxation to the ground state on a similar time scale.1,14 It is noted that transient features similar to those radical species formed from Type I reactions have been revealed in monomer solutions following two-photon ionization.4 In this work, we aim to address the dynamics of the simplest, yet most prevalent MB aggregate, MB2, in aqueous solution, which has hitherto not been investigated in detail. Aggregates of MB form face-to-face, or H-aggregates in solution, which are easily identified by their blue-shifted (hypsochromic) absorption compared with their monomer components.15−17 This observation is in agreement with molecular exciton theory, which predicts a strongly allowed transition to the upper excitonic state (S+) of a strongly coupled H-aggregate due to the addition of the in-phase transition dipole moments of constituent monomers. The lower state (S−) is dipole forbidden as a result of cancellation of the out-ofphase transition dipole combination.18 The resulting exciton states represent fully delocalized excitations in the strongcoupling limit (when Vel ≫ λ, where Vel is the electronic coupling matrix element and λ is the reorganization energy), described by electronic wave functions that are sums and differences of monomer excitations (|Ψ*A⟩/|Ψ*B⟩) leading to states completely mixed in character, i.e., |Ψ±⟩ = (|Ψ*A⟩ ± |Ψ*B⟩)/√2. The structure of MB2 is shown in Figure 1a, along
spectroscopic properties of H- and J-aggregates;19−21 however, a larger body of work has focused on J-type aggregates/ assemblies as model systems (due to their bright lower exciton manifold), while the spectroscopic properties and photophysics of H-aggregates are significantly less represented. J- and Haggregates of varying sizes serve as models for energy transfer in molecular assemblies such as nanotubes,22−25 organic semiconductors,19,21,26,27 and light-harvesting protein complexes.28−31 Given the sensitivity of electronic coupling and energy transfer rate to the orientation and distance of chromophores, these systems provide a platform to investigate energy transfer and quenching mechanisms in the nominally strong-coupling limit. In addition, covalently linked homodimers have been useful model systems of exciton dynamics since they have an intermolecular distance that is chemically well-defined.14,32−35 Coherent spectroscopy presents the opportunity to investigate exciton dynamics occurring within the first tens of femtoseconds while simultaneously resolving coherent oscillations emerging from the nonstationary superposition state (wavepacket) generated by broadband excitation. The resultant coherent dynamics yield additional information complementary to the population dynamics recovered by typical transient absorption measurements. Namely, information on the molecular structure and potential energy surfaces,36−43 and excitonic coherences along with the rate of decoherence among coupled chromophores.44−47 In particular, two-dimensional electronic spectroscopy (2DES) has been shown to be a valuable method for identifying and isolating exciton coherences directly by measuring the amplitudes of cross peaks connecting exciton states. In this work, we apply a combination of time-resolved spectroscopic methods to investigate the photophysics and nonradiative dynamics of MB and MB2 in water. The dynamics of MB have already been explored in detail;1,3,12 however, it serves as a necessary reference for examining the effects imposed on the excited state dynamics induced by aggregation. In addition, this report complements the recent vibrational coherence evaluation of MB monomer.36 Narrowband transient absorption (TA) in conjunction with target analysis was used to explore the population dynamics out to several nanoseconds with ∼100 fs time resolution. Broadband transient absorption (BBTA) and 2DES were employed to capture the early time dynamics (1 × 10−4 M) demonstrates the high degree of stabilization brought on by dispersion forces, ion−dipole, dipole−dipole, and hydrophobic interactions between MB monomers and water.16 Furthermore, the high dielectric constant of water screens the positive charges of each of the monomers, allowing for a closer approach ultimately maximizing the electronic coupling between them. Given that transitions between the lower exciton state (S−) and the ground state are forbidden, and that the rate of S+ → S− relaxation is typically fast, fluorescence is almost entirely quenched by aggregation in cases of pure H-aggregates. This is in opposition to the well-known “head-to-tail” J-aggregates whose state ordering is opposite thereby permitting fluorescence, many times with larger quantum yield than their monomers. Some studies have been undertaken to compare the
2. EXPERIMENTAL SECTION A. Experimental Methods. Methylene blue (hydrate) was purchased from Sigma-Aldrich and was dissolved in water at a concentration of ∼2 × 10−5 M for a reference sample consisting of nearly pure monomer, and ∼7 × 10−4 M to shift the equilibrium toward dimer formation. Henceforth we will refer to these two samples as “sample 1” and “sample 2” respectively. For TA, BBTA, and 2DES experiments, a 1 mm path length cuvette was used for sample 1 (dilute), yielding an optical density (OD) of 0.15. In order to avoid artifacts/line shape distortions from the high concentration sample 2,48 the path length was kept at 0.1 mm, yielding a final OD of 0.25−0.29. Steady-state absorption measurements were performed on a Cary 6000i UV−vis−NIR spectrometer, and fluorescence spectra (corrected) were taken on a Horiba PTI QuantaMaster B
DOI: 10.1021/acs.jpcb.5b11847 J. Phys. Chem. B XXXX, XXX, XXX−XXX
Article
The Journal of Physical Chemistry B 400 with slit widths of 0.38 mm for both excitation and emission. Time-resolved photoluminescence (TR-PL) measurements were done via time-correlated single photon counting on a Horiba Deltaflex spectrometer using an excitation wavelength of 653 nm from a DeltaDiode pulsed diode laser. The data were fit via reconvolution analysis with the recorded instrument response function taken from LUDOX solution. For all fluorescence measurements, an optical density of 0.05 or less in a 1 cm cuvette was used to avoid inner filter effects. For low temperature measurements, MB was dissolved in a 2:1 glycerol:water solvent mixture at concentrations similar to the dilute solution used in room temperature measurements. The sample was injected into a 0.5 mm cuvette and cooled with a liquid nitrogen coupled cryostat to 77 K. Narrowband transient absorption (TA) was carried out with a Helios transient absorption spectrometer (Ultrafast Systems) incorporating a Coherent Libra Ti:sapphire seeded amplifier operating at 1 kHz. The 800 nm output was used to pump a Coherent OPerA Solo optical parametric amplifier (OPA) for generation of the pump beam, along with generation of a white light continuum for use as the probe pulse. TA experiments were performed using white light continua in both the visible and near-infrared (NIR) regions, and the OPA pump energy was kept to ∼25 nJ/pulse. TA spectra were taken in exponentially increasing steps starting at a step size of 20 fs, and the relative polarizations of pump and probe were kept near magic angle (54.7°). Global and target analysis of the dispersion-corrected data was done using the Glotaran program.49 The spectra are presented using ΔI = Ipump − Iunpumped with ground state bleach (GSB) and stimulated emission (SE) signals positive, and excited state absorption (ESA) signals negative for direct comparison with 2DES data (which is presented in the same manner). The femtosecond experimental setup used in this work has been described in detail elsewhere,50−52 but a brief description will be given here. The 800 nm output of a Ti:sapphire seeded regenerative amplifier (Spitfire, Spectra-Physics) operating at 5 kHz was used to pump a home-built noncollinear optical parametric amplifier (NOPA) generating pulses with a bandwidth of ∼90 nm.53 The NOPA pulses were compressed using a folded grating compressor and a single prism compressor to ∼16 fs and 11−12 fs full width at half-maximum in the BBTA and 2DES spectrometers respectively, as determined via polarization-gated and transient grating frequency-resolved optical gating (PG-FROG, TGFROG).54,55 The TG-FROG profile of the NOPA pulse used in 2DES is shown in the Supporting Information. For lowtemperature 2DES experiments, the pulses remained at ∼12 fs full width at half-maximum. For BBTA, the pulses were then split into three beams by a wedged beam splitter generating two beams ∼1% of the total intensity, and used for the probe and reference pulses, thereby leaving the transmitted intensity to be used for the pump pulse. The pump was sent through a delay stage, which was scanned in 5 fs steps for the first 2 ps, after which the step size increased progressively out to 1 ns. The BBTA data shown here is the average of five separate scans, which were each averaged for 500 pump on/pump off pairs, which themselves were integrated over four pulses each. Finally, the pump energy was kept at ∼6 nJ/pulse while the probe was kept at 200−300 pJ/pulse. Excessive scatter appearing in the BBTA spectra as fringes were removed by the filtering algorithm reported previously.36
For 2DES, the compressed NOPA pulses were sent through a two-dimensional diffractive optic generating four phase-stable beams in the box geometry. Three beams (labeled E1, E2, and E3) were then passed through pairs of 1° fused silica wedges for individually controlling the time delay of each. Additionally, E3 was passed through a chopper operating at 50 Hz for active scatter/background subtraction from the data. The fourth beam was used as the local oscillator for heterodyne detection of the third-order signal, and it was fixed in time at ∼250 fs prior to the final pulse and attenuated by 104 to ensure that it did not contribute any excitations of the sample within the pulse sequence. The coherence time, t1, was scanned from −50 to +50 fs in 0.2 fs steps for sequential acquisition of nonrephasing and rephasing spectra at each waiting time, t2. The waiting time was scanned from 0 to 440 fs in 5 fs steps, and each 2D map was phased according to the projection slice theorem along with comparison of the projections of rephasing, nonrephasing, and absorptive spectra along ν1.52,56,57 The 2D spectra were generated from Fourier transformation of the data along t1 and t3 (where the t3 Fourier transform is inherently done by the spectrometer), yielding frequency-frequency maps plotted as ν1 and ν3 referring to the excitation and emission wavenumber, respectively. B. Computational Methods. All of the calculations presented here were performed using the Gaussian09 suite.58 Geometry optimizations and harmonic frequency calculations of MB2 were carried out using density functional theory (DFT) with the M05-2X functional59 employing the 6-311++G(d,p) basis set. Vertical excitation energies were calculated using timedependent DFT (TDDFT) and configuration interaction singles (CIS) with the same basis set at the ground state geometry. All of the calculations herein were performed using the polarizable continuum model (PCM) with water to incorporate nonspecific solvation effects.
3. RESULTS AND ANALYSIS A. Steady-State Spectroscopy and Time-Resolved Photoluminescence. The absorption spectra of samples 1 and 2 are shown at the top and bottom of Figure 2a, respectively, along with the fluorescence spectrum of MB taken in sample 1 with an excitation wavelength at λ = 664 nm (top, red). The dilute sample 1 yields a spectrum of effectively pure monomer, which has its absorption peak at 15060 cm−1 (664 nm) with a vibronic sub-band at 16340 cm−1 (612 nm). The corrected fluorescence spectrum displays a band at 14580 cm−1 (686 nm) with the equivalent vibronic sub-band at approximately 1300 cm−1 to lower wavenumber, indeed matching the splitting in absorption. The total Stokes shift then, using the corrected spectrum, is ∼480 cm−1. The absorption spectrum of sample 2 displays a significant increase at 16470 cm−1 (607 nm) assigned to the allowed band of MB dimer (S0 → S+). Additionally, there is some absorption that extends well to the red of the monomer band, perhaps suggesting the existence of a weak band assigned to the nominally forbidden lower exciton transition. No MB2 fluorescence that was distinguishable from the monomer could be detected, implying a significant decrease in fluorescence quantum yield upon dimerization, consistent with the expectation of static quenching from fast internal conversion to the dark exciton state. In order to disentangle the MB2 spectrum from the composite spectrum in Figure 2a (bottom), the concentrations of the monomer and dimer were calculated following the procedure of Morgounova et al. and Patil et al.15,16 With a total C
DOI: 10.1021/acs.jpcb.5b11847 J. Phys. Chem. B XXXX, XXX, XXX−XXX
Article
The Journal of Physical Chemistry B
Figure 2. (a) Absorption (black) and fluorescence spectra for samples 1 (top) and 2 (bottom) consisting of nearly pure MB and near equal amounts of MB/MB2 respectively. The laser spectrum used in BBTA and 2DES experiments is shown in blue. (b) MB (dashed red) and MB2 (blue) absorption spectra isolated from the total spectrum (black).
concentration of ∼7 × 10−4 M, the individual concentrations were found to be [MB] = 2.48 × 10−4 M and [MB2] = 2.36 × 10−4 M, and an extinction coefficient for MB2 at 607 nm was found to be ∼82000 M−1 cm−1. Using these results, the pure monomer spectrum was scaled according to its concentration and subtracted from the total spectrum yielding the dimer spectrum shown in blue in Figure 2b. Indeed, a small absorption band arises to the red at ∼14530 cm−1, presumably the result of diagonal disorder.60 Regardless, this small band can be used to approximate the exciton splitting at 2000 cm−1. Taking the reorganization energy as one-half the Stokes shift of the monomer, we then crudely approximate Vel ∼ 4λ, confirming the fully delocalized nature of the MB2 excited states. TR-PL was recorded for MB monomer, yielding an S1 lifetime of τf = 370 ± 1 ps (kf = 2.7 × 109 s−1) averaged from five independent measurements, in accordance with previous data.3,12 The results can be found in the Supporting Information. Since the cold measurements required mixing water with the cryoprotectant glycerol, the fluorescence lifetime was measured for a 2:1 glycerol:water solvent mixture at room temperature for comparison, yielding τf = 551 ± 1 ps. This result agrees with the trend of intersystem crossing rate with dielectric constant shown by Chen et al.12 Since no emission attributed to MB2 was detected, we were unable to acquire a lifetime from TR-PL. B. Transient Absorption. The transient absorption spectrum of pure MB following 664 nm excitation is shown in Figure 3a, with various slices shown in Figure 3b. The time axis in Figure 3a has been displayed in logarithmic scale for a clear representation of all time scales. Immediately following excitation, a large GSB/SE signal is observed at 14890 cm−1
Figure 3. (a) Transient absorption spectrum of MB. (b) TA spectra taken at various delay times. (c) Traces of TA signals taken at 18350 cm−1 (blue), 14860 cm−1 (green), and 11950 cm−1 (red) with fits from target analysis in black.
(671 nm), which, after ∼20 ps, has shifted down by ∼30 cm−1, demonstrating a small dynamic Stokes shift. This band peaks between the absorption and fluorescence maximum, signifying the combination of S0 → S1 GSB and S1 → S0 SE signals, and the majority of its amplitude decays predominantly with a 371 ps time constant. The signal does not decay to zero, however, leaving some residual GSB, which remains for the duration of the experiment, as indicated in the traces shown in Figure 3c. It is noted that all of the time constants given here were recovered from global fitting of the data via target analysis using the appropriate sequential model, which will be discussed in detail in the Discussion section. At higher wavenumber, several ESA bands appear at 17180 cm−1 (582 nm), 18350 cm−1 (545 nm), and a broad absorption centered around 21370 cm−1 (468 nm). All of these bands decay with nearly the same rate, suggesting that they all originate from the S1 state (S1 → Sn transitions), in agreement with the assignments made by Chen et al.11,12 This is verified by comparison with the UV transitions in the MB absorption spectrum (∼292/246 nm). To the red of the large GSB/SE signal, additional SE bands appear with much smaller amplitude. Of primary interest is the band at 12000 cm−1, which decays with the S1 lifetime, but subsequently reverses sign (Figure 3c red). This signal remains constant through 3 ns and is assigned to T1 → Tn ESA following intersystem crossing. This observation/assignment also matches that of Chen et al.12 D
DOI: 10.1021/acs.jpcb.5b11847 J. Phys. Chem. B XXXX, XXX, XXX−XXX
Article
The Journal of Physical Chemistry B
relaxation to S−, and the ESA is then assigned to S− → Sn ESA to higher lying states with overall Au or Bu symmetry. The inset in Figure 4b shows another signal unique to the aggregate sample, near the red edge of the NIR spectrum. This broad ESA signal arises instantaneously and spreads well over 3000 cm−1, the peak of which we are not capturing within this spectral range. Interestingly, all of the denoted MB2 bands decay away entirely within 20 ps, leaving only a TA spectrum of MB equivalent to Figure 3. The loss of the bleach signal indicates that the whole of the MB2 population has returned to the ground state within this period, in marked contrast to the MB monomer dynamics, which are dominated by the production of long-lived triplets. Inspection of the monomer signals in sample 2 indicates similar kinetics as the previous case (sample 1), with no indication of an increase in GSB signal as one might expect if the dimers were dissociating upon photoexcitation. BBTA was undertaken to supplement the results presented in Figure 4 for resolution of dynamics occurring within 100 fs. The parallel-polarized BBTA spectrum of sample 2 is shown in Figure 5a out to Δt = 2 ps. It is noted that magic angle measurements of MB2 resulted in equivalent kinetics as for parallel-polarization. Calculation of the anisotropy following
Proceeding forward with MB as a reference, the TA spectrum of sample 2 with 607 nm excitation is shown in Figure 4a. The
Figure 4. (a) Transient absorption spectrum of sample 2 with significant MB2 population. (b) TA spectra taken at various delay times; inset shows magnified NIR region. (c) Traces of TA signals taken at 19230 cm−1 (yellow), 16470 cm−1 (blue), 14860 cm−1 (green), and 7470 cm−1 (red) with fits from target analysis in black.
MB features are still observable as expected, but immediately striking is the stark difference in the decay rate of new features present. The most significant positive feature appears now at 16500 cm−1 (606 nm) at the MB2 bleach, and it decays drastically faster than the MB GSB/SE signal as shown in Figure 4c (blue). Such a substantial shift in kinetics already alludes to the existence of additional nonradiative channels that lead directly back to the ground state, which are not present in the monomer. The ESA signal to the blue is now peaked at ∼19200 cm−1 (520 nm) and decays with the same rate as the positive MB2 signal. The time constant for the excited state decay of MB2 is found to be 3.2 ps (k = 3.1 × 1011 s−1). No appreciable signal of the “partially-allowed” subpopulation is observed near the lower exciton transition, likely obscured by monomer signal. However, since the population is promoted to the upper exciton level, likely fast S+ → S− internal conversion takes place within the time resolution of the TA experiment, precluding detection of a dynamic shift of the S+ → S0 SE band or line shape changes. Instead, we speculate that the dominant peak at 16500 cm−1 is assigned to S+ GSB after population
Figure 5. (a) Parallel-polarized BBTA spectrum of sample 2, (b) coherent BBTA spectrum after removal of population dynamics, and (c) its Fourier transform spectrum. Dashed lines denote the wavenumber positions of the peaks of excitonic transitions in MB2 and the absorption maximum of MB. E
DOI: 10.1021/acs.jpcb.5b11847 J. Phys. Chem. B XXXX, XXX, XXX−XXX
Article
The Journal of Physical Chemistry B
Figure 6. 2D spectra at various waiting times for sample 1/monomer (left columns) and sample 2 (right columns) with absorption spectra of each above for reference.
of the total amplitude in the case of the monomer, whereas the dimer only yields ∼6%. The Fourier transform (FT) of the coherent BBTA spectrum in Figure 5c shows that many coherences of the monomer still dominate in amplitude, with frequencies going past 1600 cm−1. The same two dominant ring-stretching fundamentals at 450 and 500 cm−1 are active in MB2, with a shift in relative amplitude toward 450 cm−1, which is largest in the dimer region, implying a slight change in Franck−Condon factors upon dimerization. Despite this, however, very little vibronic activity is found to higher FT wavenumber, and the higher frequencies seen in the ESA region are effectively the same as that seen in MB. Individual traces of the oscillatory signals taken at various emission wavenumber positions are given in the Supporting Information along with their FT spectra. An active comparison between samples 1 and 2 is also given, showing very little difference in relative FT amplitude between monomer and dimer. C. 2D Electronic Spectroscopy. 2DES is an effective method for resolving dynamics with high time resolution, while still retaining spectral resolution by retrieving the excitation axis that is lost in BBTA. In addition, analysis of the full twodimensional lineshapes of spectroscopic features permits realtime observation of homogeneous dynamics between the system and the bath (vibrations and solvent). To this end, we performed 2DES on samples 1 and 2, and several 2D spectra are shown in Figure 6 at various waiting times. The results for MB (sample 1) are analogous to those reported
parallel and perpendicular polarized experiments of MB monomer (sample 1), however, yielded a rotational diffusion time of 116 ± 6 ps (2σ), and the results can be found in the Supporting Information. Power-dependent measurements of MB2 were also performed at magic angle, and no apparent differences in the measured kinetics were found. Immediately after Δt = 0, the signal near the MB2 bleach region reaches a maximum and decays rapidly in
