Photosensitizing properties of hollow microcapsules built by multilayer self-assembly of poly(allylamine hydrochloride) modified with rose Bengal

A polymeric photosensitizer based on poly(allylamine hydrochloride) (PAH) and rose Bengal (RB) was synthesized. The modified polycation PAH-RB was demonstrated to be suitable for construction of microcapsules via a layer-by-layer (LbL) assembly technique, using sodium poly(styrene sulfonate) (PSS) as counter-polyelectrolyte and CaCO3 microcrystals as templates. After CaCO3 core removal, a stable suspension of hollow microcapsules with shells incorporating RB (HM-RB) was obtained. The spectroscopic and photophysical behavior of both PAH-RB and HM-RB in aqueous environments were studied and described in terms of dye–dye interactions and dye hydrophobicity. Only HM-RB was able to generate singlet molecular oxygen with similar efficiency to free RB in air-saturated solutions upon green light irradiation. In order to explore possible practical applications as a supramolecular photosensitizer, experiments of HM-RB irradiation in the presence of chemically and biologically relevant target molecules were carried out. It was observed that is possible to use visible light to initiate the photooxidation of biological compounds in water, with many interesting advantages compared to low-molecular-weight photosensitizers such as an enhancement of the photosensitizing effect, due to a significant reduction of dye–dye interaction, or improved reuse given the straightforward size-based separation from the reaction mixture without loss of efficiency.


Introduction
Photosensitized reactions constitute one of the possible practical uses of solar radiation. In these processes, a chemical species called a photosensitizer, absorbs solar radiation in a certain spectral range and induces chemical changes in a target molecule, which does not absorb radiation. Photosensitized processes have a wide range of applications such as synthesis with controlled regio-and stereo-selectivity, 1 photodynamic therapy (PDT), 2 photoinactivation of microorganisms (PDI), 3 and wastewater treatment for pollutant removal. 4 In general, the photosensitization process may involve two types of mechanism: (i) the generation of radicals (type I mechanism, e.g., via electron transfer or hydrogen abstraction); and/or (ii) the production of singlet oxygen ( 1 O 2 ) through an energy transfer process (type II mechanism). [5][6][7] Among the main drawbacks of using homogeneous photosensitized reactions is the non-straightforward removal and reutilization procedures needed for continuous operation. As a consequence, an increasing interest in the development of strategies directed towards immobilization of photosensitizers on different platforms has been recently observed; 8,9 e.g., modication of synthetic and natural polymers with organic dyes were used in the development of new photoactive materials for water treatment, 8,10 PDI 11 and PDT. 12 Rose Bengal (RB) (Fig. 1) is a xanthene dye intensively used as photosensitizer for the 1 O 2 production, 13 due to its high solubility in water, high radiation absorption in the visible region, and high quantum yield of triplet-state production. 14 Additionally, RB can be easily anchored to different polymers using; e.g., the method introduced by Merrield, 15 which has been already employed in the development of polymer-based photosensitizers for photooxidation. [16][17][18] Photocatalytic materials based on immobilized dyes on different platforms were already reported for environmental applications. 19,20 However, the main drawback in the development of such photosensitizer materials is the tendency to form molecular aggregates driven a Instituto de Investigaciones Fisicoquímicas Teóricas y Aplicadas (INIFTA), Departamento de Química, Facultad de Ciencias Exactas, Universidad Nacional de La Plata, CCT La Plata-CONICET, La Plata, Argentina. E-mail: mariserr@inia. unlp.edu.ar by dye-to-dye interactions caused by high local concentrations, thus provoking an enhancement of non-radiative deactivation pathways and the consequent decrease of photosensitizing efficiency. 21,22 It is therefore highly desirable, to attain control over the spatial distribution of dye molecules in order to improve the design of photo-active polymeric materials. Such spatial control can be achieved for lms using layer-by-layer (LbL) assembly technique, which offers a simple and versatile construction method based on the sequential adsorption of interacting species; e.g., oppositely charged polyelectrolytes. 23 Sodium poly(styrene sulfonate) (PSS), poly(ethyleneimine) (PEI), and poly(allylamine hydrochloride) (PAH) are typical synthetic polyelectrolytes used for multilayer formation (Fig. 1). 24,25 Such lms can be conveniently functionalized via post-synthetic modication with selected molecules 26 or nanoparticles, 27,28 or by using already functionalized polyelectrolytes. [29][30][31] LbL assembly can be carried out to create lms, but also grown over different templates, such as carbonates microparticles, which aer removal with suitable chelating agents, give raise to hollow microcapsules (HM). 32 Although the use of RB in the assembly of HM was already reported, 30,33 the inuence of different anchoring strategies and its assembly into LbL architectures on their photophysical and photosensitization properties was not studied in detail. The goals of the present work can be summarized as follows: (i) to synthesize a RB modied PAH polycation (PAH-RB); (ii) to use such PAH-RB polycation for microcapsule formation via LbL with CaCO 3 microspheres as sacricial templates and PSS as counter polyelectrolyte; (iii) to evaluate the effect of polymer conformation and spatial distribution in the photophysical properties of RB; and nally (iv) to assess the photosensitizing properties of such supramolecular assembly using chemical and biological oxidizable targets.

Synthesis of PAH-RB and microcapsule preparation
The EDC/NHS mediated synthesis of PAH-RB was inspired in the bioconjugation and labelling of proteins. 34 Briey, EDC and sulfo-NHS solution was added to an RB buffered solution and stirred evenly. Then, 20 mL of PAH solution was added and stirred for 15 h. The nal product was puried by dialysis. The whole reaction and dialysis process were protected from light.
Polymeric microcapsules were built using CaCO 3 templates sequentially coated with oppositely charged polyelectrolytes layers (LbL assembly), 23,24 see Scheme 1 and ESI. † Removal of CaCO 3 template was carried by re-suspension of the coated microparticles with 0.1 M EDTA solution at pH 7 during 30 min Scheme 1 Preparation of hollow microcapsules (HM) using layer by layer self-assembly technique onto CaCO 3 template. Inset: proposed molecular structure of the PAH-RB polycation. at 25 C while shaking at 1100 rpm. A nal wash was performed using 50 mM NaCl solution to remove the remaining EDTA (Scheme 1). Finally, the obtained HM were stored at 8 C before use.
Polymer and microcapsule characterization UV-vis spectroscopy. UV-vis absorption spectra were registered with a Shimadzu UV-1800 spectrophotometer, using quartz cells of 1 cm optical path length. All the absorption spectra registered for HM were corrected using cubic polynomial extrapolations of the spectral baseline in the visible region, between 400-800 nm, as a simple approximation of scattering contribution of the sample. 35 Fluorescence emission. Steady-state and time-resolved uorescence measurements were performed using a single-photoncounting equipment FL3TCSPC-SP (Horiba Jobin Yvon), described elsewhere. 36 All the experiments were carried using 1 cm optical path quartz cells. The uorescence quantum yields (F F ) were determined from the corrected uorescence spectra using eqn (1).
where I corr is the corrected maximum intensity, A is the absorbance at the excitation wavelength (l exc ) and the superscript R refers to the reference uorophore. To avoid inner lter effects, an RB aqueous solution of absorbance below 0.1 at the excitation wavelength was used as uorescence reference (F F ¼ 0.018). The I corr values were calculated at the maximum of the uorescence band registered between 530 to 700 nm. The uorescence intensity from RB in the HM was corrected by scattering using eqn (2).
where I corr and I obs were the corrected and observed uorescence maximum intensities, A exc and A emi were the absorbance of the system at excitation and emission wavelengths, respectively. Fluorescence decays were registered with a time-correlated single photon counting (TSCPC) apparatus from Horiba. In time-resolved experiments a NanoLED source (maximum at 575 nm) was used for excitation and the emitted photons were registered by a TBX-04 detector and counted by a FluoroHub-B module. The uorescence intensity decays were tted with the Fluorescence Decay Analysis Soware DAS6® from Horiba by deconvolution of the NanoLED pulse using multi-exponential model function.
Infrared spectroscopy. Fourier-transform infrared spectra were obtained with a Jasco 4600 spectrometer with an attenuated total reectance (ATR-FTIR) accessory containing a single reection monolithic diamond crystal. Spectral resolution was 4 cm À1 and background corrections were performed by assuming an average refractive index of 1.45 for all the samples.
Dynamic light scattering and zeta-potential. All dynamic light scattering (DLS) and zeta-potential measurements were performed with a cell drive voltage of 3.3 V in 10 mM HEPES NaCl 500 mM buffer solutions at pH 7 and 25 C, using a Zetasizer Nano ZS apparatus from Malvern.
Transient absorption experiments. Laser-ash photolysis (LFP) experiments were performed using a homemade LFP system with laser pulse excitation at 532 nm (7 ns full width at half maximum, 7 mJ per pulse) as previously described. 37 The transient absorption decays of RB and PAH-RB solutions or HM suspensions were acquired with a Tektronix TDS3032B 300 MHz instrument, and up to 10 single-shot signals averaged. Prior each measurement, the solutions were degassed by Ar bubbling for 20 min. The signal analysis of the transient decays was done with the OriginPro 8.5 soware from OriginLab Corporation using a mono and/or bi-exponential tting function.
Fluorescence microscopy. Optical microscopy images were acquired on a Carl Zeiss Axio-Observer Z1 inverted microscope. Samples were illuminated using bright eld or an HBO Xe lamp complete with excitation and emission bandpass lters of 550 AE 50 nm and 605 AE 70 nm, respectively, acquired on an AxioCam camera. Zen imaging soware was used for image acquisition and analysis. All uorescence microscopy images were digitally colored post acquisition.

Steady-state photolysis experiments
Aqueous suspensions containing the target molecule and HM were irradiated in 1 cm optical path quartz cells at room temperature in continuous stirring. Two different irradiation systems were employed: (I) a Xenon arc lamp (300 W, Newport) coupled to a motorized monochromator UV-VIS (Oriel Cornerstone 130 1/8 m) by a Mounting Kit (Model 74017) and (II) a Xenon arc lamp (150 W, Luzchem) with a 475 nm lter (GG475). UV-vis absorption spectral changes of anthracene-9,10-dipropionic acid disodium salt (ADPA) were registered with a CCD-USB2000 UV-vis spectrometer (OceanOptics), with the analyzing beam placed at right angle of the photolysis beam from the irradiation setup (II).

Characterization of PAH-RB polycation
Aer covalent functionalization of PAH with RB, the resulting PAH-RB polycation was characterized thoroughly, Fig. 2 shows a comparison of ATR-FTIR absorbance spectra of RB, PAH, and PAH-RB. The carbonyl stretching mode corresponding to RB appears at 1608 cm À1 , whereas absorption peaks at 1543, 1442, and 1334 cm À1 correspond to aromatic C]C stretching modes. 38 For PAH, a signal at 1595 cm À1 corresponding to N-H in-plane bending mode appears. The spectrum obtained for PAH-RB shows bands that can be assigned to amide bond between PAH and RB; e.g., C-O stretching mode (amide I) at 1620 cm À1 , and N-H in-plane bending mode (amide II) 39 at 1535 cm À1 . Amide I and amide II bands present in PAH-RB appeared with the concomitant disappearance of carboxylic acid peak at 1608 cm À1 , indicating the amide bond formation with carboxylate group of RB. Additionally, the sharp peak corresponding to C-halogen stretching at 950 cm À1 is preserved but broadened aer functionalization.
The UV-vis absorption spectrum of PAH-RB in aqueous solutions was registered and compared with free RB (PAH is completely transparent in the visible region). PAH-RB spectrum displays one band in the green spectral region with well-dened maximum (l 1 ¼ 558 nm) and a shoulder (l 2 ¼ 523 nm), as expected for xanthene dyes (Fig. 3a). 40 However, these bands appear broader and red-shied in comparison to those of free RB. Such behavior can be associated with the self-aggregation of xanthene dyes resulting in a red shi of the absorption maxima and a decrease of the maximum-to-shoulder absorbance ratio (A 1 /A 2 ). 40 Hence, a comparison of normalized absorbance spectra of RB and PAH-RB in Fig. 3a suggests that aqueous solution of PAH-RB features the expected effect corresponding to dye-dye interactions in the polymeric structure.
The excitation/emission spectra of PAH-RB appear red-shied compared to RB (Fig. 3b and c). Such spectral shis are expected for RB in a less protic and polar environment. 41 In addition, the uorescence quantum yield (F F ) value of PAH-RB is more than one-order of magnitude lower than that corresponding to free RB (Table 1), which can be explained assuming a strong static quenching of RB uorescence as a consequence of the intense dye-dye interactions in the functionalized  polymer. In time-resolved experiments, uorescence lifetime (s F ) of RB in aqueous environments showed a mono-exponential decay with a s F 98 AE 10 ps, as expected for monomeric RB in water (see Fig. 3d). 42 Although there is some minor proportion of non-aggregated PAH-RB responsible for the observed steadystate emission, such contribution is not enough to be determined in the time-resolved experiments.
DLS analysis showed an increased hydrodynamic diameter from 122 AE 19 nm to 250 AE 50 nm, when comparing PAH and PAH-RB respectively. Considering that covalent binding of RB to PAH involve a negligible increase in average molecular weight, the observed size increment can be understood in terms of association of neighboring polymer chains caused by dye functionalization.
Zeta potential was determined for PAH and PAH-RB in aqueous solution (pH 7.0, [NaCl] ¼ 500 mM) and the values obtained were +14.0 mV and +15.0 mV, respectively. The positive value of PAH-RB is crucial for the construction of LbL architectures. The fact that the same zeta potential value was determined for both systems is consistent with the low percentage of amine groups that reacts with RB. That is, the positive charge of PAH is due to the protonated amino groups and, according to the ratio of reactants used in the synthesis of PAH-RB (see Experimental section), the resulting fraction of amino groups linked to RB molecules is very low, which means that the decrease of the total charge of the polymer is negligible.
The above discussed results conrm that synthesized PAH-RB features covalently attached RB, but also suggest the association between dye molecules on neighboring polymeric chains, which are likely to be favored by hydrophobic interactions; and cause a strong effect on the spectroscopic properties of RB.

Hollow PAH-RB microcapsules
To evaluate the integrity of microcapsules constructed via LbL with PAH-RB and PSS, optical and uorescence microscopy images were recorded before and aer removal of CaCO 3 template with EDTA. This procedure leads to the formation of hollow microcapsules, hereaer referred as HM-RB. As it can be observed in Fig. 4a and c, template removal did not signicantly change the size of microcapsules, but affected the shape registered, which was due to the loss of the rigid ller material and was expected considering previous reports. In addition, microcapsules with and without CaCO 3 core exhibit the characteristic uorescence of RB dye upon excitation at 545 nm ( Fig. 4b and d, respectively), thus conrming the presence of dye in the multilayered coating. Fig. 3a also shows the UV-vis spectrum of HM-RB aer baseline correction by scattering. RB spectrum when supported on HM features a red-shied narrower band with A 1 /A 2 ratio higher than the corresponding to PAH-RB (see Table 1). These spectral features strongly suggest that supramolecular organization aer LbL formation hinders dye-dye interactions. Accordingly, both excitation and emission spectra of HM-RB suspensions showed red-shied bands when compared to those corresponding to PAH-RB aqueous solutions ( Fig. 3a and  b). More importantly, F F of HM-RB was determined to be twenty times greater than the corresponding to dye aqueous solution (Table 1). In time-resolved experiments, the uorescence decay can be tted assuming a bi-exponential decay with an increased s F value of 786 AE 10 ps (compared to 98 ps, Table 1), together with a 27% decrease on the additional decay process, as can be observed in Fig. 3d. This complex decay for HM-RB indicates that the behavior of dye singlet excited states is strongly  inuenced by the heterogeneous environment of LbL assembly, which is in line with the observed red shi in steady state uorescence experiments (see Fig. 3c). Spectral analysis of HM-RB suggests that LbL assembly on the CaCO 3 template surface favors two effects. On the one hand the disaggregation of the dye, that is, the separation of the RB molecules, avoiding the interactions observed in PAH-RB. On the other hand the immobilization of the dye, similar to what already reported for proteins 43 and other thin lms. 44,45 It is expected that both effects hinder the deactivation pathways of the excited states of the dye, which would enhance its photosensitizing properties. This point will be analyzed in detail in the next paragraphs.
It is known that triplet excited states play a dominant role in the initiation of photosensitized processes, therefore the formation of triplet RB ( 3 RB*) species was investigated in both PAH-RB and HM-RB Ar-saturated aqueous suspensions by laser-ash photolysis experiments. In PAH-RB suspensions a monoexponential decay of 14 ms was observed, almost 23% shorter than values corresponding to free RB. Differently, for HM-RB, the measured transient absorption changed into a biexponential decay featuring longer lifetime compared to PAH-RB and similar to free RB (within experimental error, see Table 1). The above described transients can be assigned to the triplet excited state of RB in different environments; based on the observed decay rates increase in presence of O 2 . The corresponding total bimolecular quenching rate constant ðk O2 Tq Þ was calculated according to the Stern-Volmer eqn (3), and the corresponding values are reported in Table 1 (Fig. S2 †), where s 0 T and s T are the triplet lifetime of RB in the absence and in the presence of O 2 , respectively.
In this case, an increase of complexity in the assembly (from PAH-RB to HM-RB) leads to a decrease in k O2 Tq value; which can be rationalized considering that O 2 free diffusion is partially hindered in the heterogeneous assembled multilayer environment.
The results shown in this section suggest that RB molecules are accommodated into the microcapsule shell in a spatial arrangement that hinders dye-dye interaction. This supramolecular array is compatible with the observation that, upon irradiation, singlet and triplet excited states of RB ( 1 RB* and 3 RB*) are efficiently generated. The larger the local molecular rigidity around RB in the multilayer, the more favored an efficient radiative deactivation of 1 RB*; thus, conferring an intense uorescence in HM-RB suspensions. Furthermore, the formation of 3 RB* suggests that HM-RB can feature photosensitizing properties, as it is analyzed in the next section.

Photosensitizing properties of HM-RB
Although RB molecules remain immobilized into the multilayer of the HM-RB and generates 3 RB* upon irradiation, which is quenched by molecular oxygen, it is not straightforward to assume that this supramolecular system can act as an efficient photosensitizer. Generation of 1 O 2 upon irradiation of HM-RB was detected using anthracene-9,10-dipropionic acid disodium salt (ADPA) acting as chemical-trap reaction of 1 O 2 , since it becomes photo-bleached to give an endoperoxide. 46 Fig. 5a shows the UV-absorbance decreases of ADPA dissolved in the supernatant of HM-RB suspension as a function of the irradiation time using the irradiation setup I (see Experimental section). The rst-order kinetic decay of ADPA relative absorbance at 378 nm is shown in Fig. 5a inset, conrming a constant steady-state concentration of 1 O 2 . A similar signicant decrease of ADPA absorbance was observed in HM-RB suspensions, as well as in aqueous RB solutions aer 180 s irradiation (no ADPA consumption was observed in control experiments without irradiation).
In contrast, Ar-bubbling in HM-RB suspensions strongly reduces the consumption of ADPA upon illumination, suggesting that ADPA oxidation occurs by 1 O 2 -mediation to yield the corresponding endoperoxide. To conrm such assumption, oxidized ADPA solutions were heated while the absorbance was monitored in order to provoke thermolysis of endoperoxides, thus regenerating the parent aromatic anthracene ring. 47 Fig. 5b shows the recovery of typical anthracene ring absorption bands upon heating at 80 C. These results conrm the effective generation of 1 O 2 by irradiation with visible light of HM-RB suspensions.

HM-RB as photosensitizer for the oxidation of dGMP
The possible application of polymeric photosensitizers bearing RB chromophores in microcapsules was tested with the model reaction of 2 0 -deoxyguanosine 5 0 -monophosphate (dGMP) degradation as a biological target. We chose this nucleotide because in DNA guanine is the only nucleobase that reacts signicantly with 1 O 2 . 48 Therefore, aqueous suspensions of HM-RB and dGMP were exposed to UV-vis radiation using the irradiation setup II (see Experimental section). Supernatants obtained from the irradiated samples were analyzed by HPLC using the UV/vis detector. The decrease observed for peak area corresponding to dGMP was used as an indicator of nucleotide consumption in the process. In addition, several peaks with shorter retention times (t r ) were detected (Fig. 6a). It was found that HM-RB featuring 3 or 6 bilayers of polyelectrolytes can effectively photosensitize dGMP degradation (Fig. 6b). This result suggests that O 2 can diffuse across the HM-shells with different number of bilayers and interact with 3 RB* to generate 1 O 2 , that ultimately will produce the observed substrate oxidation.
Taking into account previous results, 2 0 -deoxyspiroimidantoin 5 0 -monophosphate (dSpMP) was identied as the primary product of 1 O 2 mediated reaction. 49 Analyzing the UV spectra and t r of the reaction products using irradiated HM-RB and dGMP suspension, the principal product identity can be assigned to dSpMP, which suggests the participation of 1 O 2 in the reaction mechanism. Additionally, an increase of dGMP consumption was observed as pH was increased from 7 to 9 (Fig. S3 †). It is noteworthy that dGMP chemical reactivity towards 1 O 2 is higher as pH becomes alkaline, which conrms that photosensitization mechanism of dGMP by HM-RB is 1 O 2 mediated.
The dGMP consumption aer 60 min of irradiation in successive irradiation-centrifugation cycles for HM-RB and dGMP suspensions was evaluated. For each subsequent irradiation experiment, pellets were re-suspended in dGMP fresh solutions. It is possible to reuse the same HM-RB batch since dGMP consumption was proportional to the irradiation time, and additionally, similar consumption was obtained for at least three irradiation cycles of 60 minutes each (Fig. 6b inset). This behavior showed two interesting advantages of HM-RB over RB in solution. First, the supramolecular organization prevents the dye photobleaching reactions, and second HM-RB can be easily separated by centrifugation from the reaction mixture to be reused. On the contrary, RB solution requires more sophisticated and expensive separation techniques such as chromatography.

Conclusions
In the present work we have presented a strategy for the assembly of hollow microcapsules (HM) using sacricial CaCO 3 templates, via LbL approach constituted by PSS (sodium poly(styrene sulfonate)) polyanion and a modied polycation, PAH-RB, (poly(allylamine hydrochloride)) covalently modied with rose Bengal dye. The presence of attached bulky aromatic RB molecule on modied PAH induces dye-dye interactions, presumably driven by hydrophobicity. Such aggregation modies supported RB both, spectroscopic and photochemical properties, reducing its uorescence quantum yield (F F ). The spectroscopic analysis of HM-RB revealed that the observed behavior corresponds to what can be expected for free RB in less protic and polar environments, without signicant dye-dye interactions.
Irradiation with visible light of HM-RB aqueous suspensions induces the efficient production of singlet oxygen ( 1 O 2 ) comparable to what produced by RB in aqueous solutions. Such reactive oxygen species was demonstrated to be responsible for degradation of a biological target molecule used (2 0 -deoxyguanosine 5 0 -monophosphate, dGMP). A proposed degradation mechanism is described in Scheme 2 (reactions (1)-(4)). Aer subsequent irradiation cycles and oxidation of dGMP, the reutilization of HM-RB as photosensitizer was also proved indicating that the supramolecular organization of the RB molecules into the HM shell prevents the photobleaching reactions induced by dye self-interactions.
Summarizing, we have demonstrated that by controlling the supramolecular assembly and spatial arrangement of RB attached to a polymeric structure; the photophysical and photochemical characteristics of the dye can be modulated, remarkably recovering photosensitization properties of RB in homogeneous solution with the advantage of high photostability and an easier separation for its subsequent use.

Conflicts of interest
There are no conicts to declare.