Morphologically Determined Excitonic Properties ... - ACS Publications

Jun 23, 2016 - by absorption, resonance light scattering (RLS), and resonance ... smaller total RLS intensity, compared to those of aggregates in 50 m...
0 downloads 0 Views 6MB Size
Article pubs.acs.org/JPCC

Morphologically Determined Excitonic Properties of Porphyrin Aggregates in Alcohols with Variable Acidity Christopher W. Leishman and Jeanne L. McHale* Department of Chemistry, Washington State University, Box 644630, Pullman, Washington 99164, United States S Supporting Information *

ABSTRACT: Self-assembled, excitonically coupled aggregates of 5,10,15,20tetrakis(4-sulfonatophenyl)porphyrin (TSPP) in acidified alcohols were studied by absorption, resonance light scattering (RLS), and resonance Raman (RR) spectroscopy along with scanning probe microscopy. In ethanol, the major absorption bands of aggregates in 0.5 mM HCl (sample E1) were narrower, with smaller total RLS intensity, compared to those of aggregates in 50 mM HCl (sample E2). RR scattering cross sections were smaller, and depolarization ratio dispersion greater, for E2 than E1. Two distinct types of aggregates at differing acidity in 1-propanol, analogous to the ethanolic samples, but only one type in methanol, were also identified. Atomic force microscopy (AFM) images for E1 showed small, single layered structures (∼5−20 nm diameter, ∼1.5−2 nm thickness). Variable morphologies for E2 included double-layered, elongated structures consistent with collapsed nanotubes (∼25 nm width, ∼3.5−4 nm thickness, ∼100−200 nm length) and single molecule thick sheets (∼50−200 nm in characteristic diameter). The former were also observed in ultrahigh vacuum scanning tunneling microscopy (UHV-STM) images. Aggregate morphology appears to depend on protonation and deprotonation kinetics of porphyrin monomers. Spectroscopic observations suggest that a larger subset of J-band excitonic transitions are significantly active in E2 than in E1. These variations in excitonic properties are attributed to the differing aggregate sizes, shapes, and possibly molecular packing. Formation of long nanotubes appears to require a solvent able to donate two hydrogen bonds, such as water. An open sheet structure is favored otherwise, as in ethanol.



INTRODUCTION Natural photosynthetic systems have inspired extensive research into artificial mimics to harness excitonic processes for light harvesting.1−4 Polymer and other thin film organic photovoltaics (OPV) are analyzed largely in terms of excitonic processes.5,6 The sizes and distributions of ordered and disordered regions determine excitonic energy transfer and charge separation in both natural and artificial light-harvesting systems.7,8 More detailed understanding of the interplay of these structural and dynamical properties is needed in order to rationally design systems for improved performance. To this end, model systems for which processing conditions can be controlled to reproducibly tune the relevant structural parameters are indispensable. Self-assembled 5,10,15,20tetrakis(4-sulfonatophenyl)porphyrin (TSPP) aggregates are such a system and have the added benefit of being wellcharacterized in reference conditions (aqueous HCl at concentration ≳0.1 M).9−12 In recent work, we correlated counterion-dependent morphologies of TSPP aggregates with variations in excitonic absorption and resonance light scattering (RLS) spectral features.13 From resonance Raman (RR) spectroscopy, we identified effects of nanotube bundling and disorder on the relative strengths of transitions of differing polarization and coupling between such states, both in solution and in individual aggregates.14 Here, we examine aggregates prepared in acidified alcohols to explore other dimensions of © XXXX American Chemical Society

our general hypothesis that solution environment shapes aggregate morphology, which in turn determines excitonic properties. In the reference conditions of aqueous HCl, the tetra-anionic free base species (Soret band λmax = 413 nm) predominates at pH ≳ 5.5 (Figure 1a, with none of the color-coded protons). Decreasing pH, two protons (Figure 1a, red) add to the porphyrin core nitrogens to form the diacid dianion, which exhibits a Soret band λmax at 434 nm and an enhanced Q-band with λmax = 665 nm. For pH ≲ 3, protonation of two of the peripheral sulfonate groups (Figure 1a, two of the blue protons) converts the diacid dianion monomer to the tetraacid dizwitterion, which aggregates via pairing between cationic porphyrin cores and anionic sulfonate groups. The aggregate absorption spectrum shows a Soret band split into a blueshifted “H-band” (λmax ≈ 420 nm) and a red-shifted and narrowed “J-band” (λmax = 490 nm) as well as a red-shifted and further-enhanced Q-band (λmax = 698 nm).9−12 At HCl concentrations above about 4 M, protonation of the remaining sulfonate groups produces hexa-acid dication monomer (Soret band λmax = 413 nm; Figure 1a, the other two blue protons).15 Received: May 17, 2016 Revised: June 21, 2016

A

DOI: 10.1021/acs.jpcc.6b04998 J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C

protonation steps occur in alcohols differ from those in water, so these experiments may provide clues to details of aggregate internal structure. One possible such structure is depicted in Figure 1b for a small aggregate fragment. HCl-induced aggregates in aqueous solution are primarily nanotubes with typical lengths from a few hundred nanometers to several micrometers, diameters ∼15−20 nm, and wall widths ∼1.5−2 nm (single molecule thickness).18−22 For oriented aggregates, the J- and H-bands and the red and blue sides of the J-band exhibit different degrees of parallel or perpendicular electronic polarization relative to the aggregate axis.11,17,23 Aggregates induced with rapid, unidirectional stirring or with chiral solution additives show helicity in circular dichroism spectra,11,24,25 and cryogenic transmission microscopy has revealed the nanotube wall to be composed of 26 helical chains.26 Excitation frequency dependence (dispersion) of RR depolarization ratios indicates that the J-band is composed of multiple, closely spaced transitions.21,22 Larger RLS and smaller RR intensity in D2O/DCl than in H2O/HCl indicated greater spatial coherence in the former and suggested that water intimately associates with the aggregate structure and plays a role in excitonic coherence.22 These results highlighted the significance of hydrogen bonding between porphryin and solvent molecules. We recently investigated how perturbations to hydrogen bonding due to alkali ions in aqueous solution affect aggregate structure,13,14 and for aggregates in alcohols, we expect hydrogen bonding to influence aggregate structure quite differently than in water. For example, while a single water molecule could plausibly bridge any of the sulfonate oxygens on two neighboring porphyrins, a single alcohol molecule can only bridge protonated sulfonate oxygens. Salts can induce TSPP aggregation without protonation of sulfonate groups.11,18,27−30 We recently reported counteriondependent variations in TSPP aggregate morphology, with corresponding variations in absorption, RLS, and RR spectra.13,14 We attributed these variations to weak excitonic coupling between bundled nanotubes and to different degrees of disorder in bundling and in transition dipole moment orientations within nanotubes. NaCl and KCl promoted bundling of nanotubes. NaCl-induced bundles had apparently more rigid nanotubular units and were markedly more orderly than those induced by KCl. The exciton delocalization as measured by RLS was relatively greater in the presence of NaCl. Relatively greater strength of excitonic transitions polarized perpendicularly to the aggregate long axis in both NaCl- and KCl-induced aggregates was indicated by smaller ratios of low to high frequency mode RR intensities and larger depolarization ratios compared to the mostly nonbundled HCl-, LiCl-, and CsCl-induced aggregates. We proposed that weak excitonic coupling between nanotubular subunits preferentially enhanced transversely polarized excitonic transitions. CsCl-induced aggregates were mostly individual nanotubes, showing a slightly lower RLS intensity (and thus lower coherence) compared to the HCl induced aggregates. We attributed this to disorder in transition dipole moment orientations related to looser molecular packing in the presence of CsCl. Aggregates induced in alcohols provide examples allowing a different perspective on relationships between morphology and excitonic properties. In contrast with water, alcohols donate a maximum of one hydrogen bond per molecule rather than two and have hydrophobic alkyl groups. They are thus sometimes viewed as archetypal amphiphiles. Whereas our previous work

Figure 1. (a) TSPP monomer, with colors highlighting pairs of protons corresponding to transformation between different monomer species. (b) Two views of a small aggregate fragment. On the right, red and blue arrows represent monomer transition dipole moment components contributing primarily to the J- and H-bands, respectively. (c) Total extinction spectra, concentration-scaled to equivalent of 5 μM aggregated TSPP, for samples with 0.75 M HCl in water (black; sample W), 0.5 mM HCl in ethanol (magenta; sample E1), and 50 mM HCl in ethanol (cyan; sample E2).

It is uncertain whether the first two sulfonate groups to be protonated are in opposite or adjacent locations on the tetraphenylporphyrin (TPP) skeleton. STM images of dizwitterionic TSPP monolayers suggest that opposite positioning is favored,16 but the aggregate structure has been modeled assuming adjacent positioning, based on other imaging evidence.17 The locations of neutral sulfonic acid and anionic sulfonate groups should affect the orientations of monomers about aggregate lattice sites. Acid concentrations at which these B

DOI: 10.1021/acs.jpcc.6b04998 J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C

Cross sections for the 859 and 1034 cm−1 modes of 1-propanol and methanol, respectively, were calculated by an external standard method. These were used as secondary internal standards for sample cross sections of aggregates in those solvents. The total RR scattering cross section σRR,s for a mode in the sample (or in the secondary standard, in the case of methanol and 1-propanol solvents) was calculated from the reference cross section σRR,r using the equation34

looked primarily at effects of hierarchical assembly of nanotubular subunits into bundles, here we focus on how excitonic properties depend on the internal structure of primary aggregate units. Both entropic effects and steric hindrance lead to a solvent hydrogen bonding structure dominated by chains, in contrast to the highly interconnected networks typical of water.31 These differences in hydrogen bonding should result in different solvation structures around the aggregates in alcohols. We will consider possibilities such as formation of a micelle-like alcohol envelope around aggregates which could influence aggregate size and shape.32 In this work, we compare absorption, RLS, and RR data, along with AFM images, among various samples of TSPP aggregates formed in alcohols, to develop further insight into morphology−excitonic structure correlations. Distinct primary aggregate structures, identified in solutions at different acidity, enable exploration of such correlations on a scale intermediate between the local molecular packing and higher levels of hierarchical structure. In this way, we illuminate the role of solvent−porphyrin hydrogen bonding in fundamental aggregate unit structure and resulting variations in excitonic coherence and coupling between transitions of differing polarization.

σRR,s =

nr 2crIs ⎛ 1 + 2ρs ⎞⎛ 1 + ρr ⎞ ⎜⎜ ⎟⎟⎜⎜ ⎟⎟σRR,r ns 2csIr ⎝ 1 + ρs ⎠⎝ 1 + 2ρr ⎠

where nr,s are the refractive indices of the reference and sample solvents, cr,s are the concentrations of the reference and sample, Ir,s are the measured intensities of the reference mode in the standard and of the mode of interest in the sample, and ρr,s are the corresponding depolarization ratios. The RR total scattering cross section σRR,r of the reference mode pertains to ethanol external standard for the methanol and 1-propanol calculations or to the alcohol used in each particular sample. Ir,s were calculated as the sum of polarized and depolarized integrated peak areas after spline fit background subtractions. We assume that at these dilute TSPP concentrations the sample refractive indices are equal to those of the pure solvents, so we ignore that factor for internal standard calculations. Cyclohexane (spectrophotometric grade, Acros) was used as a frequency standard to calibrate the wavenumber scale each time the monochromator gratings were moved. All data processing for RR experiments was performed in Origin 2015. AFM Imaging. Samples were prepared by depositing a 10 μL drop of each solution on a mica disc and allowing to air-dry. Images were acquired in tapping mode, with the tip driven at resonant frequency of 338 Hz, on a Veeco Dimension scanning probe microscope under ambient atmospheric conditions, as described elsewhere.13 UHV-STM Imaging. Substrates of Au(111) on mica were prepared by published methods,35 H2 flame-annealed, and stored in a vacuum desiccator prior to use. Samples were prepared by depositing a 10 μL drop of solution on the substrate and allowing to air-dry. Images were acquired in constant current mode on an Omicron low temperature ultrahigh vacuum scanning tunneling microscope (LT-UHV STM) at the Environmental Molecular Sciences Laboratory of Pacific Northwest National Laboratory, Richland, WA. The sample was held at a temperature 80 K with pressure of ∼3 × 10−11 Torr. Bias voltage of +2.0 V and tunneling current of 100 pA were used to produce the images shown in this publication.



EXPERIMENTAL METHODS Sample Preparation. TSPP free base monomer solutions at concentrations of 50 μM in ethanol and in methanol and 20 μM in 1-propanol were prepared from TSPP tetrasodium salt (Frontier Scientific, Inc., Logan, UT). Ethanol (200 proof), 1propanol, and methanol were all 99.9%+ purity. 1 M HCl solutions were prepared by diluting concentrated aqueous HCl stock solution with each alcohol. For each sample HCl concentration, a portion of the 1 M solution was diluted with the appropriate alcohol to twice the final desired concentration. An equal volume of 10 μM TSPP in the same alcohol was added to the acid solution, and the 5 μM TSPP samples were allowed to equilibrate for 2−3 days in the dark prior to data acquisition. We selected several primary samples which showed a maximum degree of aggregation (minimal residual monomer absorption). These are designated as E1 (0.5 mM HCl in ethanol), E2 (50 mM HCl in ethanol), P1 (0.2 mM HCl in 1propanol), P2 (20 mM HCl in 1-propanol), and M (40 mM HCl in methanol). The mole percent of water in these samples was approximately 0.01% in E1, 1% in E2, 0.005% in P1, 0.5% in P2, and 0.55% in M. UV−Vis Extinction, Absorption, and RLS Spectroscopy. Extinction spectra were acquired on a Shimadzu UV−vis spectrophotometer. Each sample was thoroughly vortexed to disperse flocculated aggregates just prior to data acquisition. For the primary samples (E1, E2, P1, P2, and M), scattering spectra were acquired with a PTI QuantaMaster spectrofluorometer with xenon lamp excitation, in synchronous scan mode at 0.25 nm increments, ∼2 nm excitation, and ∼0.125 nm emission bandpass. Extinction spectra and scattering spectra were mutually corrected to produce “true” absorption and RLS spectra, using calibration and data analysis codes in Matlab, as described previously.13 Resonance Raman Spectroscopy. Polarized resonance Raman data were collected in confocal backscattering geometry, and integrated intensities and depolarization ratios calculated from the spectra after subtracting the fluorescence background, as described elsewhere.14 Here we also calculated absolute RR scattering cross sections. Published total cross section values for the 880 cm−1 mode of ethanol were used as reference data.33



RESULTS Variations of Extinction Spectra with HCl Concentration. Total extinction (absorption plus scattering) spectra for TSPP solutions in ethanol with varying HCl concentration, scaled to the free base monomer absorption maximum, are shown in Figure 2: 0−0.5 mM HCl (Figure 2a), 0.5−50 mM HCl (Figure 2b), and 50−500 mM (Figure 2c). Similar series of spectra for TSPP solutions in 1-propanol and methanol are shown in Figure S1. All aggregate spectra are consistent with structures in which the angles of one Soret band component of each nearest-neighbor monomer pair is less than 54.7° with respect to the core-to-core direction, yielding a negative excitonic coupling that gives rise to the J-band, peaking at 480−495 nm, depending on the sample. The other monomer Soret band component is at 90° with respect to this direction, C

DOI: 10.1021/acs.jpcc.6b04998 J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C

At the smallest HCl concentrations in ethanol (Figure 2a), free base begins to convert to diacid dianion (λmax = 438 nm). By about 0.1 mM HCl, a small amount of aggregate is evident, becoming the predominant species at 0.5 mM HCl, with a Jband peak at 486.5 nm. With further increase in HCl concentration (Figure 2b), the three major aggregate bands all shift slightly toward the red (see Table 1). J- and Q-band peak extinction values increase slightly as aggregation reaches its maximum extent at 50 mM HCl (J-band λmax = 492.5 nm), having broadened from about 380 to 960 cm−1. This species disappears as HCl concentration is increased beyond 50 mM (Figure 2c). The spectrum begins to resemble that of diacid dianion monomer in aqueous solutions of pH 3−4, although all bands are red-shifted (Soret λmax = 444.5 nm) with respect both to aqueous solution (434 nm) and the diacid dianion monomer (438 nm) in lower HCl concentrations in ethanol. This is identified as the fully protonated monomer. For aggregates in 1-propanol, we observe a similar pattern: first, diacid dianion forms in coexistence with free base, and then a first aggregate species which predominates around 0.2 mM HCl (Figure S1a, J-band λmax = 485 nm). This species diminishes in prevalence, while the spectral bands broaden slightly and red-shift, reaching a second maximum J-band intensity at 20 mM HCl (Figure S1b, λmax = 489.5 nm). These effects are not as pronounced as in ethanol. Fully protonated monomer forms with further increase in acid concentration (Figure S1c). The aggregation behavior of TSPP as a function of acidity in methanol differs from that in ethanol and 1-propanol (Figures S1d,e). Here we identify only one distinct aggregate species, for which the maximum prevalence occurs at 40 mM HCl (J-band λmax = 492.5 nm). Notably, this aggregate is always less abundant than monomer. Once again, fully protonated hexa-acid dication forms at high HCl concentration. Absorption spectra of the fully protonated monomer in each alcohol are shown in Figure S1f. Absorption and RLS Spectra of Distinct Aggregate Types. Two distinct aggregate types are prevalent at different HCl concentrations. The shorthand designations and corresponding solution conditions are given in the Experimental Methods. Absorption and RLS spectra for E1 and E2 are shown in Figure 3. Corresponding data for samples P1, P2, and M are shown in Figure S2. On each plot of absorption or RLS data, the reference sample with 0.75 M HCl in water (designated sample W) is also shown. The spectrophotometric and scattering data have been mutually corrected for scattering and self-absorption, respectively, and scaled using the residual monomer band so that the spectra represent equivalent concentrations of aggregated TSPP. Characteristic quantities for the J-band in absorption and RLS spectra of each sample are displayed in Table 1.

Figure 2. Variation of extinction spectra with HCl concentration for 5 μM TSPP in ethanol. Spectra are scaled to the free base monomer absorption maximum: (a) no acid to 0.5 mM HCl, (b) 0.5−50 mM HCl, and (c) 50−500 mM HCl.

resulting in positive coupling that gives rise to the H-band, with a peak in the 420−425 nm region.12 The aggregate Q-band, λmax ≈ 700 nm, is thought to derive its shape from excitonic coupling within the Q-band as well as vibronic effects.36 Table 1. J-Band Spectroscopic Data samplea

absorption λmax (nm)

absorption fwhm (cm−1)

J/H-band separation (cm−1)

RLS λmax (nm)

integrated RLS intensity (au)

C̃ (Abs) rel

C̃ (RLS) rel

W E1 E2 P1 P2 M

490.0 486.5 492.0 485.0 489.5 492.5

362 384 902 724 721 605

3290 3020 3240 2770 3150 b

490.0 487.5 496.0 490.5 490.5 496.5

602 53.8 435 10.0 132 442

≡1 0.89 0.40 0.25 0.25 0.36

≡1 0.30 0.85 0.13 0.47 0.86

a

Sample conditions defined in text. bH-band location in sample M is obscured by the residual monomer Soret band. D

DOI: 10.1021/acs.jpcc.6b04998 J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C

Figure 3. Concentration-scaled absorbance (a) and RLS (b) spectra for samples W, E1, and E2. Concentration-scaled total extinction, absorption, and RLS extinction spectra for samples E1 (c) and E2 (d).

absorption line width does not follow a simple 1/√Ncoh dependence expected for a band arising from a single electronic transition.37,38 For comparison, we show the absorption-based = wAbs2[W]/wAbs2[sample], where coherence parameter C̃ (Abs) rel wAbs is the full width at half-maximum (fwhm) of the absorption J-band. As the integrated RLS intensity provides a more reliable lower bound on the overall coherence of the J-band, we take as the appropriate quantity for comparing the extent of C̃ (RLS) rel excitonic delocalization. The lack of correlation between C̃ (Abs) rel and C̃ (RLS) shows that the J-band is composed of multiple rel transitions and that this composite band structure depends on for E2 and M are similar solution conditions. Values of C̃ (RLS) rel and are closer to that for W than those for the other samples. for P2 is intermediate, while those for E1 and P1 are C̃ (RLS) rel both significantly smaller. The magnitude of J/H-band peak separation in the absorption spectra is approximately equal to the excitonic coupling strength and expected to depend on the number as well as the arrangement of coupled monomers. The coupling strength thus estimated from absorption spectra follows the same order among the samples as C̃ (RLS) rel . Unlike the J-bandwidth, the coupling strength should not depend strongly on the composite band structure. The consistency of these two rankings thus shows qualitative agreement between signatures of excitonic coherence in absorption and RLS spectra, despite the failure of the absorption line width to scale accordingly. TSPP Aggregation Processes in Alcohols. Figure 4 shows successive extinction spectra during the first few hours of aggregation in solutions E1 and E2. Spectra for the free base monomer (dark red) and the diacid dication (dashed gray) are shown for reference. After mixing ethanolic 1 mM HCl and 10 μM TSPP free base solutions in equal proportions, the E1 solution first shows quick (