Spontaneous self-assembly of water-soluble porphyrins having poly(ethylene glycol) as branches: Dependence of aggregate properties from the building block architecture

Chemical Physics 409 (2012) 23–31

Contents lists available at SciVerse ScienceDirect

Chemical Physics

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

Spontaneous self-assembly of water-soluble porphyrins having poly(ethyleneglycol) as branches: Dependence of aggregate properties from the buildingblock architecture

Valentina Villari a,⇑, Placido Mineo a,b,⇑, Emilio Scamporrino b, Norberto Micali a

a CNR-IPCF Istituto per i Processi Chimico-Fisici, Viale F. Stagno d’Alcontres 37, I-98158 Messina, Italyb Dipartimento di Scienze Chimiche, Università di Catania, Viale Andrea Doria 6, I-95125 Catania, Italy

a r t i c l e i n f o a b s t r a c t

Article history:Received 14 April 2012In final form 22 September 2012Available online 2 October 2012

Keywords:PorphyrinLight scatteringAggregated structures

0301-0104/$ - see front matter � 2012 Elsevier B.V. Ahttp://dx.doi.org/10.1016/j.chemphys.2012.09.022

⇑ Corresponding authors.E-mail addresses: [email protected] (V. Villari), gmi

In the present paper, conformational and structural properties of aggregated structures, spontaneouslyformed via ‘‘self-assembly’’ in water solution, were studied by means of static and dynamic light scatter-ing, UV–vis and Fluorescence Spectroscopy measurements.

It was observed that small changes on the chemical structure [different position of poly(ethyleneglycol) hydrophilic branches around the porphyrin unit] or molecular size (with the presence of two orthree porphyrin units inside a cyclic structure) of the starting compound resulted in changes of the opti-cal properties, size and even local architecture of the self-assembled aggregates. In particular thepoly(ethylene glycol) branches position around the porphyrin cores influenced significantly the forma-tion of H- or J-type porphyrin aggregates and their properties.

� 2012 Elsevier B.V. All rights reserved.

1. Introduction

Natural bio-systems are often used as models for the design ofnew molecules or supramolecular species for applications in manyapplied fields as ‘‘new materials’’, ‘‘nanotechnology’’, ‘‘pharmaceu-tical science’’, etc. [1]. Among the bio-processes, the spontaneousmolecular self-assembly, generating well-ordered structures fromdisordered components, is certainly one of the most fascinatingand studied phenomenon. Just to mention, the self-assembly ofamphiphilic structures [2] was the keystone of the formation ofthe primordial organized cell-based life [3,4].

In this context, multi-porphyrin structures play a relevant rolein many biological processes, like, for example, in the circulararrangement of the bacteriochlorophyll-a chromophores in rhodo-pseudomonas acidophila, allowing the conversion of solar energyinto chemical energy [5,6]. More generally, the strong electroniccoupling and excitation energy localization [7–9], occurring inporphyrin aggregates, [10–21] are responsible for their relevantproperties, useful for potential applications in the mimesis ofphotosynthetic centers as well as for application in solar cells, non-linear optics, electronic devices, etc. [22].

Porphyrin-based systems are very versatile both from the syn-thesis and the supramolecular point of view. Indeed, differentmolecular architecture can be obtained via synthetic procedures

ll rights reserved.

[email protected] (P. Mineo).

whereas final aggregate forms with proper structural and opticalproperties can be controlled and tuned through self-assembly. Inmany cases, porphyrin water solubility can be also guaranteed bythe presence of charged groups in the porphyrin peripheralposition.

These systems join the properties of porphyrins in interactingwith many chemical species [15,23–28] and the possibility to con-trol their self-assembly by varying properly the thermodynamicparameters of the solution, like pH, ionic strength or concentration[29–32].

In the self-assembly of charged porphyrin systems, the forma-tion of specific aggregates (H- or J-type) originates from the ionicinteractions between porphyrin peripheral charges and the proton-ated pyrroles of other porphyrin molecules [33].

Unfortunately, models based on charged porphyrins are notsuitable for mimicking natural systems because, generally, the lat-ter are constituted of uncharged porphyrins (e.g. bacteriochloro-phyll-a, in rhodopseudomonas acidophila). Furthermore, thepresence of charges on porphyrins determines a drastic change oftheir spectroscopic properties (i.e. modification of the UV–visabsorption spectra and/or fluorescence quenching).

Despite the importance of modeling natural antenna systems, itis still quite rare to find studies on uncharged porphyrin- [34] ordye-derivatives [35,36] in water solution.

To study the aggregation properties of uncharged porphyrins,we synthesized new water soluble covalent porphyrins with twohydrophilic polyethyleneglycolic (PEG) branches bound in two

24 V. Villari et al. / Chemical Physics 409 (2012) 23–31

different porphyrin peripheral positions (compounds P-cis and P-trans, Fig. 1a and b); two cyclic oligo-porphyrins [c(P-cis)m], deri-vated from P-cis (Fig. 1c), were also investigated. The balance be-tween the hydrophobic porphyrin core and the hydrophilic PEGschains affected their aggregation phenomena. The obtained datawere compared with those of a previously studied porphyrin, hav-ing four polyethylene glycolic branches in peripheral positions [31](P-Star, Fig. 1d). This last system have shown interesting colloidalproperties, forming rigid aggregates arranged in fractal structuresthrough a Diffusion Limited Aggregation (DLA) mechanism [31,37].

In the present paper, conformational and structural propertiesof aggregated structures, spontaneously formed via ‘‘self-assem-bly’’ in water solution, were studied by means of Static and Dy-namic Light Scattering, UV–vis and Fluorescence Spectroscopymeasurements.

It was observed that small changes of the chemical structure(different position of the PEG hydrophilic branches around the por-phyrin unit) or molecular size (dimeric or trimeric cyclic porphyrinform) of the starting compound resulted in changes of the opticalproperties, size and even the local architecture of the self-assem-bled aggregates.

In particular the position of the two poly(ethylene glycol)branches around the porphyrin cores (Fig. 1a and b) influenced sig-nificantly the formation of H- or J-type porphyrin aggregates andtheir properties.

2. Experimental section

2.1. Materials

2.1.1. Synthesis of 5,15-di[p-(x-methoxypolyethylenoxy)phenyl]-10,20-di[p-hydroxyphenyl] porphyrin (P-trans) and 5,10-di[p-(x -methoxypolyethylenoxy)phenyl]-15,20-di[p-hydroxyphenyl]porphyrin (P-cis)

P-cis and P-trans (Fig. 1a and b) with an average molecular massof about 2100 g/mol and a narrow polydispersity (Mw/Mn = 1.01),were prepared by reaction between tetrakis(p-hydroxyphenyl)por-phyrin (in turn obtained from pyrrole and p-acetoxybenzaldheydein boiling propionic acid [38]) and chlorinated poly(ethylenegly-col)methyl ether having an average molecular mass of about

Fig. 1. Structure of the investigated porphyrins: (a) P-cis; (b) P-trans; (c) c[P-cis]m with mp�17, for a, b and c, and p�9 for d.

750 Da [obtained by reaction of poly(ethyleneglycol) mono methylether with thionyl chloride in tetrahydrofurane (THF)] [39]. Theobtained products, dissolved in CHCl3, were fractionated by a chro-matographic column, using silica gel as stationary phase and asolution of CHCl3/C2H5OH/N(C2H5)3 (96,5/2,0/1,5) as eluant. Asidentified from MALDI-TOF and 1H-NMR analyses (see SupportingInformation), the third and the fourth colored compounds elutedfrom the column were P-trans and P-cis, respectively.

2.1.2. Synthesis of di-{5,10-di[p(x -methoxypolyethylenoxy)phenyl]-15,20-di [p-hydroxyphenyl]porphyrin} [c(P-cis)2] and tri-{5,10-di[p(x-methoxypolyethylenoxy)- phenyl]-15,20-di[p-hydroxyphenyl]porphy-rin} [c(P-cis)3] cyclic formals

[c(P-cis)2] and [c(P-cis)3], with the two or three porphyrin unitslinked by means of methylene bridges (Fig. 1c), were synthesizedby interfacial etherification reaction between P-cis and a large ex-cess of dibromemethane in a Toluene/H2O mixture, in presence ofNaOH and tetrabuthyl ammonium-bromide (used as a phase-transfer agent), under stirring and at refluxing temperature. Thematerial contained in the organic phase was dried by rotoevapora-tion in vacuum and c[P-cis]2 and c[Pcis]3 isolated by Gel Perme-ation Chromatography. The collected fractions were characterizedby MALDI-TOF and 1H NMR analyses (data not shown for brevity).

2.1.3. Synthesis of 5,10,15,20-tetrakis-[p-(x -methoxypolyethylenoxy)phenyl]porphyrin (P-star)

The synthesis of P-star (Fig. 1d), here used for comparison, wasmade according a previous work [31]. As an example, 0.5 g of PEG-MEC-350 (about 1.43 mmol) were dissolved in 6 mL of a H2O/THF(1/1) solution; after the addition of 0.12 g of tetrakis(p-hydroxy-phenyl)-porphyrin (0.177 mmol, half of the stoichiometric re-quired quantity) dissolved in 1.42 mL of a 0.5 M NaOH aqueoussolution, the mixture was refluxed 24 h. Then, 1.42 mL of 0.5 MNaOH and 1.5 mL of THF were added and the solution continuouslyrefluxed.

After further 24 h, the reaction was stopped by the addition ofCH3COOH, the mixture was dried under vacuum, and the residue,dissolved in CH3COCH3, fractionated by column chromatogra-phy using silica gel as stationary phase and a solution ofCHCl3/C2H5OH/N(C2H5)3 (97/2/1) as eluant. Several fractions,

= 2 and m = 3, respectively; (d) P-star. The average polymerization degree of PEG is:

V. Villari et al. / Chemical Physics 409 (2012) 23–31 25

constituted of porphyrin derivatives with a different number ofpolyethylenoxy branches, are recovered. The yield of the P-star(first eluted fraction) was about 30% with respect to the initialporphyrin amount.

2.2. Methods

2.2.1. Light-scattering measurementsFor the dynamic and static light-scattering experiments at wide

angle (range 20�–150�) a He–Ne (k = 633 nm, at a power of 10 mW)and, for the P-cis solution, a Nd:YAG (k = 532 nm, at a power of100 mW) laser sources, with the light linearly polarized orthogonalto the scattering plane and adopting a homemade computer con-trolled goniometric apparatus, were used [40]. The higher powersource at k = 532 nm was necessary because of the low scatteredintensity of the P-cis solution, whereas for the other solutionsthe source at k = 633 nm was preferred for the low extinction coef-ficient (absorbance lower than 0.05). However, the absorbance ofP-cis solution at 532 nm was about 0.05 and then not affectingthe scattering data. To distinguish between polarized and depolar-ized scattering contributions, a Glan-Thomson analyser was placedin the scattered beam. The depolarization ratio, defined as ‘‘IVH/(IVV � 4/3IVH)’’ (with the subscripts VV and VH referred to the polar-ized and depolarized intensity), provided information about theoptical anisotropy of the building blocks of the aggregate. Thedepolarization ratio was 60.03 for all the solutions at 633 nmand equal to 0.004 for P-cis solution at 532 nm (see Table 1); there-fore, the contribution of the rotational motions in the measuredpolarized correlation function (see the Appendix) was calculatedas less than 4% for all the solutions investigated. Due to the noisydepolarized correlation functions measured, the data analysiswas carried out only for the polarized correlation function, neglect-ing the contribution from rotations.

The dynamic light-scattering experiments at small angle (range3�–8�) were performed in a modulated hetero-beating (instead of aself-beating) configuration, in order to improve the detection of thescattered light. Due to the small scattering angles, no polarizationanalysis was carried out. For this experiment the He–Ne laser beamwas split before impinging on the sample with a portion passingthrough the sample and the other one used as reference, a phasemodulated at 300 Hz and then recombined with the scatteredbeam. It can be considered that, for the solutions investigated,the concentration was low enough that the interaction amongaggregates can be neglected and the diffusion coefficient identifiedwith that at infinite dilution. Moreover, only a single relaxationprocess was detected for all the samples and the polydispersity,as obtained by the cumulant analysis described in the Appendix,was about 0.2.

The static light-scattering experiments at small angle (SALS)were performed with an apparatus with a Fourier lens to focusthe scattered wave front on a linear CCD detector (a fibered 1024element diode array). In order to collect only the scattered portionof the light, the detector was set off the transmitted beam. A 16 bitanalog-to-digital converter was employed to digitalize the signal.The intensity scattered by the solvent was taken as the background

Table 1Summary of the measured hydrodynamic radius and of the structural and conformational pcis and the correlation length of the other aggregates, depending on the model used for thethe less concentrated solutions.

Sample RH (lm) R or n (lm) Aggr

P-cis 0.12 0.2 HP-trans 0.37 0.8 Jc[P-cis]2 0.8 (0.2) >5 (3.5) Jc[P-cis]3 0.55 (0.25) >5 (1.5) –

and subtracted from all the measured spectra. After the geometri-cal corrections (due to the refraction at the flat cuvette surface),SALS data were matched with SLS data by normalization with asecondary standard. However, in the fitting procedure of the inten-sity profile, a free multiplicative parameter (close to unity) was in-serted in order to take into account the unpolarized SALS geometryand, hence, to improve the matching of the two data sets. Temper-ature was set at 25 �C and controlled within ±0.01 �C.

2.2.2. Spectrophotometric measurementsUV–visible spectra were recorded on a Shimadzu UV-1601

spectrophotometer at T = 25 ± 0.1 �C using quartz cuvettes with1 cm path length and THF or water as a solvent. The deconvolutionof the spectra was performed by using Gaussian band shape in thefrequency domain.

Excitation spectra were measured by means of a Varian CaryEclipse Fluorescence spectrophotometer using cuvettes with 1 cmpath length.

2.2.3. Gel permeation chromatographyA PL-GPC 110 Polymer Laboratories thermostated system,

equipped with two PL-gel 5 Mixed-D and one Mixed-E 5 columnsattached in series, was used to isolate c[P-cis]2 and c[P-cis]3. Theseparation was performed using THF as an eluant at T = 25.0 ±0.1 �C and a flow rate of 1 mL/min. Both a differential refractometer(Polymer Laboratories) and a UV–vis spectrophotometer (HP1050), connected in parallel, were used as detectors.

2.3. Samples preparation

The investigated concentrations in water were: 6.8 lM, for P-cis; 2.8 lM, for P-trans; 0.45 mM and 5 mM, for c[P-cis]2; 0.6 lMand 10 lM, for c[P-cis]3; 4 lM, for P-star.

The choice of the right concentration for each of the investi-gated macromolecules was crucially subordinated to the corre-sponding light-scattered intensity, which had to be high enoughto get a good signal-to-noise ratio; so that also the extinction spec-tra were collected at those concentrations. The solutions were pre-pared by dissolving the solid in pure distilled water, waiting 24 hbefore performing measurements. For each experiment, the solu-tions were prepared and characterized twice to ensure the repro-ducibility of the results.

3. Results

For a more accurate study, the aggregation behavior of porphy-rins was examined both in water and organic solvents. The Soretregion of the absorption spectra of P-cis, P-trans and P-star solu-tions in THF, centered at 421 nm, shown in the inset of Fig. 2, dis-played the typical features of a porphyrin macrocyclic solution[41]. Furthermore, the very similar profile of the curves empha-sized a no-dependence on the chemical structure. Instead theUV–vis spectra of the corresponding aqueous solutions, shown inFig. 2, exhibited significant differences with the appearance of blueand red shifts.

roperties of the aggregated forms. The third column reports the measured radius for P-fitting. The values in parenthesis for the c[P-cis]2 and c[P-cis]3 measured radii refer to

egate type Rigidity Dep. ratio 633 nm (532 nm)

No 0.005 (0.004)Yes 0.03 (0.06)No 0.025 (0.035)No 0.009 (0.004)

Fig. 2. Extinction spectrum of P-cis (trace a), P-trans (trace b), P-star (trace c), c[P-cis]2 (trace d) in water and in THF (in the inset).

26 V. Villari et al. / Chemical Physics 409 (2012) 23–31

In principle, several mechanisms could give rise to a red shiftphenomenon: (i) a protonation of the porphyrin core [42,43]; (ii)a bathochromic effect due the solvent [44,45]; (iii) a flattening ofaromatic group in meso positions [46]; (iv) an aggregation phe-nomenon [47]. In the present case, the protonation of the porphy-rin core was excluded because the experiments were performed ina neutral medium. Also the solvent effect on the Soret band wasconsidered unlikely because, generally, it is very small (about25 nm). Concerning the possibility of an aromatic substituenttwisting and flattening (observed when the porphyrin is close toa planar surface [46] to form, as an example, a flattening supramo-lecular complex), it was not considered, in the present case, sincethe large polyether groups attached to the phenyl moieties shouldprevent the rotation.

As a consequence, the blue and red shifts, here observed, wereattributed to the exciton splitting originated from the formationof H- and J-type aggregates [33,48,49].

For the P-cis aqueous solution, the shape of curve a (Fig. 2) wasconsidered consistent with a exciton splitting of the B transition,caused by the molecular aggregation; in particular, by means ofthe deconvolution of the UV-signal (shown in supporting informa-tion), the decrease of the amplitude of the monomer band at422 nm, in favor of both a weaker red-shifted band at about439 nm and a blue-shifted band at 405 nm, was evidenced. Ofthese signals, the dominant contribution at 405 nm was attributedto a local arrangement of the porphyrin macrocycles with a face-to-face alignment of the transition moments [50,51] compatiblewith H-aggregates (as sketched in Fig. 3) rather than a head-to-tailalignment (J-aggregate).

On the contrary, the more intense band at 440 nm, observed inthe P-trans (trace b) and P-star (trace c) cases, was more consistentwith a J-aggregation. Then the PEG-branches played a crucial rolein the aggregate structure formation.

The splitting energy of the neighboring porphyrin units in anaggregate can be explained by applying the exciton coupling the-ory [52,53]. By indicating the excitonic state of a dye aggregate(constituted by N two-level molecules) with |W(k)> and the interac-tion Hamiltonian with H = H0 + V (H0 being the unperturbed Ham-iltonian and V the perturbation operator), it follows that thecoupling interaction between two neighbouring molecules in theaggregate is Vnm = <n|V|m>. In the dipole–dipole approximation,the interaction operator between singly excited states is expressedin terms of the dipole moment operator, ~l. Then, the interactionenergy between two dipole moments in the aggregates is

Vnm ¼~ln �~lm � 3ð~ln � r̂nmÞð~lm � r̂nmÞ

4pe0n2r3

where ln ¼ hWð0Þn j~lnjWðkÞn i is the transition dipole moment of the

molecule n in the aggregate between the kth excited state and theground state, e0 the permittivity of free space, n the refractive indexof the medium, r the center-to-center chromophore distance andr̂nm the corresponding unit vector. By assuming that the monomertransition dipole moment, l0, is not affected by the interactionswith other monomers in the aggregate and that the angle betweentransition dipole moments is zero (ideal H- and J-type aggregates),it is

V ¼ l20

4pe0n2r3 ð1� 3 cos2 #Þ ð1Þ

where # is the tilt angle between the transition dipole moments inthe aggregate and the vector r̂nm. V is negative for the J-aggregatered-shifted band and positive for the H-aggregate blue-shifted one.

The splitting energy for larger arrays of N porphyrins, deter-mined by the observed energy difference between the blue or thered-shifted band and the monomer band, is described by the Eq. (2)

DE ¼ 2Vcosðp=ðN þ 1ÞÞ ð2Þ

Typically, for N larger than 20 molecules the exciton couplingtheory predicts that extinction of the aggregate band increasesand its bandwidth decreases, whereas its wavelength keeps almostthe same. This means that, with a good approximation, the energyshift depends only on the nearest-neighbor interactions. As it willbe discussed later, the self-assembly phenomenon in water gener-ated large aggregates which can be reasonably considered to con-sist of more than 20 units. Therefore, the difference in the splittingenergy among P-cis, P-trans and P-star could be attributed to acombination among the different distance between porphyrins,the monomer transition dipole moment strength and the tilt angle,rather than to different aggregation number. In fact, the transitiondipole moment of the three monomers was the same in THF solu-tion (as resulted from the trace a, b and c in the inset of Fig. 2) butdifferent in water as a consequence of the presence of the phenolic-OH groups and the molecule symmetry. The formation of J-typeaggregates, for both P-trans and P-star, indicated that the brancheslength of the latter are still short enough not to hinder the head-to-tail approach of the porphyrin cores. Indeed, it was observed [28]that only when the polymerization degree of all the four arms in-creases to about 17, extended aggregation does not take place.

Even though in porphyrin aggregates the angle between transi-tion dipole moments could be different from zero (varying the ratiobetween the red and blue-shifted band amplitude), the predomi-nant contribution of one of the bands can suggest, at least qualita-tively, that the different hindrance of the PEG arms in the threemolecules could be responsible for the different local porphyrinarrangements. The UV–Vis spectrum in THF of the planar c[P-cis]2 (a cyclic dimer of P-cis, trace d in the inset of Fig. 2) showed,as expected, two evident exciton splitted transitions at 417 and424 nm. The profile in water (curve d) was more complex resultingin two main bands at 392 and 450 nm, respectively, and two lessintense bands at about 423 and 441 nm. The features of thespectrum indicated the characteristic exciton splitting of theaggregated form of c[P-cis]2, with the appearance, in particular,of the narrow red-shifted J-band at 450 nm with respect to thebis-porphyrin band at 423 nm (see the sketch in Fig. 3).

The architecture of the cyclic c[P-cis]3 along with a certain de-gree of distortion of each porphyrin ring from the plane (seeFig. 4) could be responsible for the single broad [54] Soret bandat 422 nm observed for its solution in THF (spectrum not shown).More interesting, it was the loss of shape of the absorption profilein water (inset of Fig. 5) except for a huge scattering contribution

Fig. 3. Sketch of the H- and J-type porphyrin arrangement for some of the investigated macromolecules, obtained by optimization with the MM+ method. Polyethylene glycolbranches are not shown.

Fig. 4. Structure of c[P-cis]3 obtained by optimization with the MM+ method.Polyethylene glycol branches are not shown.

Fig. 5. Excitation spectrum of c[P-cis]3 in water at 10 lM. The arrows indicate thepeaks of the deconvolved bands. In the inset the corresponding exctinctionspectrum is reported.

V. Villari et al. / Chemical Physics 409 (2012) 23–31 27

indicating an extended aggregation. Therefore, the electronic com-munication among the cyclic porphyrin trimers in the aggregatesmust be weak, likely for the reduction of their approaching by ste-ric hindrance. Due to this weak electronic communication, it waspossible to exploit the fluorescence emission at 658 nm by whichwas collected the excitation spectrum shown in Fig. 5. The decon-volution procedure for the main band, around 420 nm, showed anenvelope of three bands (almost equivalent) corresponding,

respectively, to the non-aggregated species (at 425 nm) and toboth H-type (at 404 nm) and J-type (at 440 nm) aggregates. Prob-ably, the lack of planar architecture could have caused a more dis-ordered local structure, so affecting the tilt angle betweenchromophores and, consequently, the H-/J-type band ratio.

In order to investigate to what extent a different local arrange-ment of the aggregate building blocks could affect the structureand the dynamic properties of the aggregates themselves at a mes-oscopic level, static and dynamic light-scattering measurementswere carried out.

Fig. 7. Absolute scattered intensity profile for c[P-cis]2 in water at 5 lM (circles)and for c[P-cis]3 in water at 10 lM (squares). In the inset the absolute scatteredintensity at small angle for c[P-cis]2 in water at 0.45 lM is reported.

Fig. 8. Decay rate for c[P-cis]2 at 0.45 lM (squares) and for P-trans at 2.8 lM(diamonds), as examples. The dashed lines indicates the Q2-dependence.

28 V. Villari et al. / Chemical Physics 409 (2012) 23–31

The scattered intensity from the P-cis aggregates in water waslower than that of the P-trans and, despite a similar slope at thehighest exchanged vector values Q, its angular profile, displayedin Fig. 6, appeared to bend at lower Q. This occurrence, strength-ened by the absence of a detectable scattered signal at small angle(Q < 5 lm�1), seemed to indicate that P-cis formed smaller aggre-gates. However, the Q values investigated were still too large(QR > 1) so that an estimation of the gyration radius through a Gui-nier fit was not reliable. The fit of the P-cis scattering profile with aform factor of a polydisperse sphere gave a mean radius of about200 nm (see inset of Fig. 6). On the other hand, P-trans aggregatesclearly possessed a fractal structure, with a correlation length, n, ofabout 0.8 lm (and Rg =

p3n �1.4 lm) and a fractal dimension

Df = 2.5 ± 0.1, as indicated by the fit with the Chen-Texeira relationin Fig. 6 (see Appendix). The latter recalled the structural proper-ties of P-star in water reported in reference [31] but the larger sizeof the P-trans aggregates was responsible for the higher slope athigh Q region (surface fractal). As far as c[P-cis]2 and c[P-cis]3 wereconcerned, they formed very large aggregates with fractal struc-ture (see Fig. 7), the fractal dimension being Df = 2.5 ± 0.1. The lackof the bending at low Q suggested that those aggregates had a ra-dius higher than 5 lm. All these fractal structures seemed to indi-cate a mechanism of growth typical of the Diffusion-LimitedAggregation (DLA) [55–57].

A smaller aggregate size, as an example, was obtained at lowerconcentration, as indicated in the inset of Fig. 7 for c[P-cis]2; in thiscase, in fact, the Ornestein–Zernike law (described in Appendix)was applicable at small Q values giving a correlation length ofabout 3.5 lm.

As far as the dynamical properties of the aggregates were con-cerned, the angular dependence of the correlation function decayrate (see Appendix) clearly indicated that, besides possessing a dif-ferent mesoscopic arrangement, the aggregates had differentdynamical properties. In particular, the decay rate of the polarizedcorrelation function of P-cis displayed a Q3-dependence as a func-tion of the exchanged wave vector, whereas that of c[P-cis]2 and ofc[P-cis]3 displayed two regimes. These last appeared in Fig. 8, inwhich the polarized decay rate of the correlation function of c[P-cis]2 and P-trans were reported, as an example. For c[P-cis]2 {aswell for c[P-cis]3} the quadratic Q-dependence C = DQ2 was ful-filled only at low Q (where QR < 1); in this region it was possibleto obtain the translation diffusion from which the hydrodynamicradius was then calculated (see Table 1 and Appendix). At high

Fig. 6. Absolute scattered intensity profile for P-cis in water at 6.8 lM (squares) andfor P-trans in water at 2.8 lM (circles) for the whole Q range at small and wideangle with the fit result with the Chen-Texeira relation. The inset reports theintensity profile for P-cis along with the fit according to the polydisperse sphereform factor.

Q, corresponding to QR > 1, the upward bending followed aQ3-dependence, indicating the existence of internal motions. Acrossover from Q2- to Q3-dependence of the decay rate was absentfor P-trans aggregates (Fig. 8). This occurrence indicated that, un-like P-cis and its dimer and trimer species, P-trans aggregates (asalso reported for P-star aggregates in reference [31]) despite large,were rigid. It could be also noticed that the internal mobility wasretained also for the smaller aggregates of c[P-cis]2, obtained atlower concentration, whereas the smaller aggregates of c[P-cis]3

obtained at lower concentration, appeared to be still rigid.These features were summarized in Fig. 9 in which the Q-

dependence of the correlation function decay rate was normalized[58] by the size of the aggregates. The crossover region at QR � 1appeared here more clearly. As a matter of fact, by inspectingFig. 9, the Q3 regime was not well fulfilled by the hCi values ofthe larger aggregates of c[P-cis]3; rather, an intermediate regimewas reached. By exploiting the crossover from Q2 to Q3 atQRH = 1, it was possible to extract an estimation of the hydrody-namic radius also for the P-cis, for which, due to the very low scat-tering, the dynamic scattering apparatus at small angle did notfurnish correlation function with a good signal-to-noise ratio. Byscaling the hCi/Q2 values of P-cis in order that the data superim-pose to the Q3-dependence plotted in Fig. 9, the diffusion coeffi-cient (and hence the corresponding hydrodynamic radius) was

Fig. 9. Normalized decay rate for all the investigated moieties in solution: P-cis at6.8 lM (stars), P-trans at 2.8 lM (diamonds), c[P-cis]2 at 0.45 lM (full squares) andat 5 lM (open squares), c[P-cis]3 at 0.6 lM (open circles) and at 10 lM (full circles).The two dotted lines indicate the Q2-and Q3-dependence of the decay ratecorrelation function.

V. Villari et al. / Chemical Physics 409 (2012) 23–31 29

extracted. The calculated hydrodynamic radius of about 120 nmseemed to be consistent with the radius obtained by the static lightscattering measurements.

4. Discussion

The investigated macromolecules were properly designed tostudy the effect of the chemical architecture on their self-assembly.Although P-cis and P-trans monomers had a very similar averagemolecular mass, their chemical structure was different for thediverse position of the two PEG hydrophilic groups around theporphyrin core (indeed for P-star the four PEG arms were shorter,as described in the caption of Fig. 1). The occurrence that the por-phyrin cores of P-cis molecules took a face-to-face arrangement(H-type) inside the aggregates, was likely due to the steric hin-drance of PEGs and to the more effective self-shielding of thehydrophobic cores.

It might be hypothesized that the internal structure of P-cisaggregates was not homogeneous and that, analogously to otherstructures investigated, it could be fractal-like, but, due to thesmall size, this issue could not be clarified by the results.

In P-trans, instead, the opposed position of the PEG arms, madethe macromolecule more symmetric appearing the head-to-tailarrangement to involve a minor steric hindrance of the PEG chains,as resulted from the direction of preferential growth along one ofthe transition moments (J-type building blocks). In this case, thecore surface exposed to the solvent was so wide that the strongerp–p interaction contribution drove towards the formation of largefractal aggregates. This case displayed many analogies with P-star,even if the presence of the –OH groups in P-trans may have playedan important role in increasing the strength of the exciton couplinginteraction.

Stacking phenomena due to p–p interactions were more pro-nounced in c[P-cis]2, due to the larger region occupied by the pla-nar bis-porphyrin core. On the other hand, c[P-cis]3 could give riseto more complex arrangements both because each single porphy-rin core, in principle, could be exciton coupled with one of the por-phyrin cores of other molecules and because, due to its non-planararchitecture, could form non-vanishing tilt angles between thecores.

The internal mobility of the aggregates should surely be ledback to the local conformation of the constituting building blocks

but not to their stacking features. In other words, both P-transand c[P-cis]2 aggregates could be constituted by J-type buildingblocks, but not all with a rigid structure. This lack of aggregaterigidity could be ascribed to the steric hindrance caused by thelong PEG chains, which could hamper the close approach of theporphyrin cores.

Finally, the building blocks of these porphyrin-based macro-molecules were able to depolarize the light, in agreement withthe results reported in reference [29], in which the anisotropy inthe polarizability tensor was attributed to the strong exciton cou-pling between the porphyrins in the aggregate. In that case, how-ever, only J-type aggregates were formed. Furthermore, bycomparing the different local structure originated by the differentmolecular architectures, it turns out that the depolarization washigher for J-type than for H-type arrangement.

5. Conclusions

The results here reported show how the structural and confor-mational properties at a mesoscopic level can be modulated by aproper design of the molecular architecture. Both the stacking phe-nomenon and the steric hindrance contribution as well as therigidity of the aggregates, can be varied by changing the peripheralpositions of the PEG branches in the porphyrin macrocycle. More-over, the steric hindrance of PEG chains seemed to affect also thelocal structure of the aggregate building blocks and the electroniccoupling between adjacent porphyrin cores. The possibility to tunethe optical properties, the size and even the local structure of theself-assembled aggregates, opens many perspectives for applica-tions of these or similar compounds in material science andnanotechnology.

Acknowledgments

This work was supported by MIUR PRIN 2008 projects no.2008KHW8K4, 20088NTBKR and PON 2010 project DIATEME(PON01_00074).

Appendix A

Light. scattering data analysis

Dynamic. light scatteringThe scattered light was collected in a pseudocross correlation

mode through two cooled R943–02 photomultipliers at the samescattering angle [59]. Only the scattered light impinges onto thedetector (self-beating mode) and it is analyzed by a MALVERN4700 correlator to build up the normalized intensity autocorrela-tion function [60,61]

g2ðQ ; tÞ ¼hIðQ ;0ÞIðQ ; tÞihIðQÞi2

ð3Þ

where Q is the exchanged wavevector, whose absolute value is|Q| = (4pn/k) sin(h/2) (h being the scattering angle, n the refractiveindex of the solution and k the wavelength of light in vacuum).For scattered electric fields obeying Gaussian statistics, the Siegert’srelation holds

g2ðQ ; tÞ ¼ 1þ ajg1ðQ ; tÞj2 ð4Þ

where a is a constant depending on the experimental setup and g1-

(Q,t) = hES⁄(Q,0) ES (Q,t)i/hI(Q)i is the normalized scattered electricfield autocorrelation function. For optically anisotropic particles likeporphyrin aggregates (the anisotropy in the polarizability tensorderives from the strong exciton coupling between porphyrins)[29], the polarized scattered electric field autocorrelation function

30 V. Villari et al. / Chemical Physics 409 (2012) 23–31

[60,62,63], under the condition of independence of translationaland rotational motions and for dilute solutions, can be written as

gVV1 ðQ ; tÞ ¼

1hIVV ðQ ;0Þi

NPðQÞFðQ ; tÞ a2 þ 445

b2FrðtÞ� �

ð5Þ

where N is the number of polymers constituting the aggregate, P(Q)the form factor of the aggregate, a the isotropic excess polarizabilityof the aggregate with respect to the solvent, and b the anisotropy ofthe aggregate polarizability. From a general point of view, F(Q,t)contains information on the aggregate translational and internalmotions and Fr(t) is a self rotational correlation function that de-pends only on the reorientation of the aggregate (independent of Q).

For diffusing monodisperse optically isotropic spherical scatter-ers the normalized scattered electric field autocorrelation functiontakes a simple exponential form, g1(Q,t) = exp(�C � t). Under thecondition QR� 1, C is related to the diffusion coefficient, D, bythe relation C = DQ2. However, if the diffusing spherical scattereris rigid, the latter relation is fulfilled also for QR > 1.

For polydisperse scatterers, the field autocorrelation functionmay be expressed as the Laplace transform of a continuous distri-bution G(C) of decay rates, each related to differently sized diffus-ing entities: g1(Q,t) =

RG(C)exp(�C � t) � dC;. In the case of

monomodal distribution of decay rates, the mean diffusion coeffi-cient hDi = hCi/Q2 (where hCi =

RC G(C)dC is the mean decay rate)

can be obtained from the standard cumulant analysis [60,61]

ln jg1ðQ ; tÞj ¼ �hCit þ 1=2!l2t2 � 1=3!l3t3 þ � � � ð6Þ

with ln the moments of the distribution G(C). The polydispersityindex is related to the second moment l2 (variance) as l2/hCi2(for values lower than 0.3). At infinite dilution, the Einstein-Stokesrelation, RH = kBT/(6pgD), can be used to obtain the hydrodynamicradius RH of the scatterers from their diffusion coefficient (kB beingthe Boltzmann’s constant, T the absolute temperature and g the sol-vent viscosity).

For very large and non-rigid aggregates, the QR < 1 region wasinvestigated by dynamic light scattering experiments at small an-gle. Information on C was obtained by the analysis of the powerspectrum of the scattered electric field, which, for the Wiener–Kintchine theorem is the time Fourier transform of its correlationfunction (i.e. of g1(t)) and has the form [64]

IðxÞ ¼ hIip

C

C2 þx2ð7Þ

By setting the value of the reference beam intensity at least oneorder of magnitude higher than the scattered one, the power spec-trum was measured. Under this condition, it is given by Eq. (7), butcentered at the modulation frequency [61,64] and the half-width athalf maximum directly gives C.

Static. light scatteringFrom the static light scattered intensity measurements, after

subtracting the solvent contribution and normalizing by the inten-sity of a reference sample (toluene), the absolute excess scatteredintensity is obtained [60,61,65]

IðQÞ ¼ KMwcPðQÞSðQÞ ð8Þ

P(Q) and S(Q) being the normalized form factor and structurefactor, respectively, Mw the molecular weight of the scatterer, cthe mass concentration and K the optical constant [61].

Mathematical expressions have been derived for particles of dif-ferent shapes [60,66,67]. For homogeneous spherical particles withradius R, for instance, it is

PðQÞ ¼ 3

ðQRÞ3sinðQRÞ � ðQRÞ cosðQRÞ½ �

" #2

ð9Þ

For polydisperse spheres the form factor is obtained as an aver-age weighted by the size distribution function.

For large aggregates whose density correlation function obeysan exponential law with decay rate n (representing the correlationlength), the scattered intensity profile can be described by theOrnestein–Zernike form

IðQÞ / ð1þ Q 2n2Þ�1 ð10Þ

For large scatterers possessing a self-similar structure, fractals,for which the spatial correlation density fluctuation is a homoge-neous function, the scattered intensity obeys the following powerlaw

IðQÞ / Q�Df ð11Þ

where Df is the fractal dimension. This scaling law is typical of frac-tally arranged ideal systems. Real fractal systems have a finite sizeand the Chen and Teixeira [68] structure factor takes into accountboth the power law behaviour at higher Q values and the Orn-stein–Zernike form at lower Q values in the following relation

IðQÞ / sin½ðDf � 1Þ arctanðQnÞ�ðDf � 1ÞQnð1þ Q 2n2ÞðDf�1Þ=2 ð12Þ

with n the cut-off correlation length.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.chemphys.2012.09.022.

References

[1] G. Whitesides, M. Boncheva, PNAS 99 (2002) 4769.[2] J.-M. Lehn, Angew. Chem. Int. Ed. Engl. 29 (1990) 1304.[3] W. Fox, Biosystems 2 (1968) 235.[4] D.W. Deamer, J.P. Dworkin, Topics in Current Chemistry (Prebiotic Chemistry)

259 (2005) 1.[5] A. van Oijen, M. Ketelaars, J. Köhler, T. Aartsma, J. Schmidt, Chem. Phys. 247

(1999) 53.[6] J. Chambron, V. Heitz, J-P Sauvage, in: K.M. Kadish, K.M. Smith, R. Guilard

(Eds.), The Porphyrin Handbook, Academic Press, vol. 6; p 1.[7] T. Kano, H. Kobayashi, J. Chem. Phys. 116 (2002) 184.[8] T. Ogawa, E. Tokunaga, T. Kobayashi, Chem. Phys. Lett. 410 (2005) 18.[9] E. Collini, C. Ferrante, R. Bozio, J. Phys. Chem. B 109 (2005) 2.

[10] E.G. McRae, M. Kasha, in: L. Augenstein, R. Mason, B. Rosenberg (Eds.),Physical Processes in Radiation Biology, Academic Press, New York, 1964, p.23.

[11] O. Ohno, Y. Kaizu, H. Kobayashi, J. Chem. Phys. 99 (1993) 4128.[12] N.C. Maiti, M. Ravikanth, S. Mazumdar, N. Periasamy, J. Phys. Chem. 99 (1995)

17192.[13] J.M. Ribò, J. Crusats, J.A. Farrera, M.L. Valero, J. Chem. Soc., Chem. Commun.

(1994) 681.[14] D.L. Akins, H.R. Zhu, C. Guo, J. Phys. Chem. 98 (1994) 3612.[15] R.F. Pasternack, E.J. Gibbs, P.J. Collings, J.C. de Paula, L.C. Turzo, A. Terracina, J.

Am. Chem. Soc. 120 (1998) 5873.[16] R. Rubires, J. Crusats, Z. El-Hachemi, T. Jaramillo, M. Lopez, E. Valls, J.A. Farrera,

J.M. Ribò, New J. Chem. 23 (1999) 189.[17] P.J. Collings, E.J. Gibbs, T.E. Starr, O. Vafek, C. Yee, L.A. Pomerance, R.F.

Pasternack, J. Phys. Chem. B 103 (1999) 8474.[18] S.C.M. Gandini, E.L. Gelamo, R. Itri, M. Tabak, Biophys. J. 85 (2003) 1259.[19] M. De Napoli, S. Nardis, R. Paolesse, M. Vicente, R. Lauceri, R. Purrello, J. Am.

Chem. Soc. 126 (2004) 5934.[20] N. Micali, F. Mallamace, A. Romeo, R. Purrello, L.M. Scolaro, J. Phys. Chem. B

104 (2000) 5897.[21] M. Castriciano, A. Romeo, V. Villari, N. Angelini, N. Micali, L. Scolaro, J. Phys.

Chem. B 109 (2005) 12086.[22] A.D. Schwab, D.E. Smith, B. Bond-Watts, D.E. Johnston, J. Hone, A.T. Johnson,

J.C. de Paula, W.F. Smith, Nano Lett. 4 (2004) 1261.[23] T. Kurtàn, N. Nesnas, F.E. Koehn, Y. Li, K. Nakanishi, N. Berova, J. Am. Chem. Soc.

123 (2001) 5974.[24] V.V. Borovkov, N. Yamamoto, J.M. Lintuluoto, T. Tanaka, Y. Inoue, Chirality 13

(2001) 329.[25] A. Mammana, G. Pescitelli, T. Asakawa, S. Jockusch, A. Petrovic, R. Monaco, R.

Purrello, N.J. Turro, K. Nakanishi, G.A. Ellestad, M. Balaz, N. Berova, Chem. Eur.J. 15 (2009) 11853.

V. Villari et al. / Chemical Physics 409 (2012) 23–31 31

[26] V. Villari, P. Mineo, N. Micali, N. Angelini, D. Vitalini, E. Scamporrino,Nanotechnol. 18 (2007) 375503.

[27] N. Angelini, N. Micali, V. Villari, P. Mineo, D. Vitalini, E. Scamporrino, Phys. Rev.E 71 (2005) 021915.

[28] N. Angelini, N. Micali, P. Mineo, E. Scamporrino, V. Villari, D. Vitalini, J. Phys.Chem. B 109 (2005) 18645.

[29] N. Micali, V. Villari, M. Castriciano, A. Romeo, L.M. Scolaro, J. Phys. Chem. B 110(2006) 8289.

[30] M. Castriciano, A. Romeo, V. Villari, N. Micali, L.M. Scolaro, J. Phys. Chem. B 107(2003) 8771.

[31] N. Micali, V. Villari, P. Mineo, D. Vitalini, E. Scamporrino, V. Crupi, D. Majolino,P. Migliardo, V. Venuti, J. Phys. Chem. B 107 (2003) 5095.

[32] V. Crupi, R. Giordano, D. Majolino, P. Migliardo, V. Venuti, N. Micali, V. Villari, P.Mineo, D. Vitalini, E. Scamporrino, Mol. Phys. 101 (2003) 1517.

[33] F. Würthner, T.E. Kaiser, C.R. Saha-Möller, Angew. Chem. Int. Ed. 50 (2011) 3376.[34] C. Frixa, M.F. Mahon, A.S. Thompson, M.D. Threadgill, Org. Biomol. Chem. 1

(2003) 306.[35] T. Miyatake, H. Tamiaki, M. Fujiwara, T. Matsushita, Bioorg. Med. Chem. 12

(2004) 2173.[36] T. Miyatake, H. Tamiaki, Coord. Chem. Rev. 254 (2010) 2593.[37] V. Crupi, D. Majolino, P. Migliardo, V. Venuti, N. Micali, V. Villari, P. Mineo, D.

Vitalini, E. Scamporrino, J. Mol. Struct. 651 (2003) 675.[38] E. Scamporrino, D. Vitalini, Macromol. 25 (1992) 1625.[39] P. Mineo, D. Vitalini, E. Scamporrino, Macromol. Rapid Commun. 23 (2002) 681.[40] V. Villari, N. Micali, J. Pharm. Sci. 97 (2008) 1703.[41] S. Matile, N. Berova, K. Nakanishi, J. Fleischauer, R. Woody, J. Am. Chem. Soc.

118 (1996) 5198.[42] S.S. Cady, T.J. Pinnavaia, Inorg. Chem. 17 (1978) 1501.[43] A. Gulino, P. Mineo, S. Bazzano, D. Vitalini, I. Fragalà, Chem. Mater. 17 (2005)

4043.[44] Z. Chernia, D. Gill, Langmuir 15 (1999) 1625.[45] V.G. Kuykendall, J.K. Thomas, Langmuir 6 (1990) 1350.[46] Y. Xu, L. Zhao, H. Bai, W. Hong, C. Li, G. Shi, J. Am. Chem. Soc. 131 (2009) 13490.[47] G. de Miguel, M. Perez-Morales, M.T. Martin-Romero, E. Munoz, T.H.

Richardson, L. Camacho, Langmuir 23 (2007) 3794.

[48] R. Pasternack, C. Bustamante, P. Collings, A. Giannetto, E. Gibbs, J. Am. Chem.Soc. 115 (1993) 5393.

[49] A. D’Urso, R. Randazzo, L. Lo Faro, R. Purrello, Angew. Chem. Int. Ed. 49 (2010)108.

[50] N. Aratani, A. Osuka, H. Cho, D. Kim, J. Photochem. Photobiol. C 3 (2002) 25.[51] J. Parkash, J.H. Robblee, J. Agnew, E. Gibbs, P. Collings, R.F. Pasternack, J.C. de

Paula, Biophys. J. 74 (1998) 2089.[52] M. Kasha, H.R. Rawls, M. Ashraf El Bayuomi, Pure Appl. Chem. 11 (1965) 371.[53] M. Castriciano, A. Romeo, V. Villari, N. Micali, L.M. Scolaro, J. Phys. Chem. B 108

(2004) 9054.[54] Y. Nakamura, I.-W. Hwang, N. Aratani, T. Ahn, D.M. Ko, A. Takagi, T. Kawai, T.

Matsumoto, D. Kim, A. Osuka, J. Am. Chem. Soc. 127 (2005) 236.[55] H.E. Stanley, N. Ostrowsky (Eds.), Growth and Form: Fractal and Non–Fractal

Patterns in Physics, Martinus Nijhoff Publishers, Dordrecht, 1986.[56] F. Mallamace, N. Micali, Riv. Nuovo Cimento 15 (1992) 1.[57] T. Vicsek, Fractal Growth Phenomena, World Scientific, London, 1992.[58] S. Magazù, G. Maisano, F. Mallamace, N. Micali, Phys. Rev. A 39 (1986) 4195.[59] F.T. Arecchi, M. Corti, V. Degiorgio, S. Donati, Opt. Commun. 3 (1971) 284.[60] B.J. Berne, R. Pecora, Dynamic Light Scattering, Wiley-Interscience, New York,

1976.[61] B. Chu, Laser Light Scattering–Basic Principle and Practice, Academic, San

Diego, 1991.[62] R. Piazza, V. Degiorgio, M. Corti, J. Stavans, Phys. Rev. B 42 (1990) 4885.[63] M.M. De Souza Lima, J.T. Wong, M. Paillet, R. Borsali, R. Pecora, Langmuir 19

(2003) 24.[64] H.Z. Cummins, H.L. Swinney, in: E. Wolf (Ed.), Progress in optics, vol. 8,

Amsterdam, North-Holland, 1970, pp. 135–200.[65] F. Mallamace, N. Micali, in: W. Brown (Ed.), Light Scattering: Principles and

Development, Oxford Science Publications, 1996, p. 381.[66] B. Chu, Z. Zhou, in: I. Baianu, H. Pessen, T. Kumosinski (Eds.), New Techniques

and Applications of Physical Chemistry to Food Systems, vol. 2, Van NostrandReinhold, New York, 1993, p. 245.

[67] O. Glatter, L. Kratky, Small Angle X-ray Scattering, Academic Press, New York,1983.

[68] S. Chen, J. Teixeira, Phys. Rev. Lett. 57 (1986) 2583.