Singlet oxygen phosphorescence photosensitized in nano-aggregates of C₆₀ buckminsterfullerene is insensitive to solvent and quenchers and strongly red-shifted indicating highly polarizable interior

Abstract

We investigated the strongly red-shifted singlet oxygen (¹O₂) phosphorescence spectra in an aqueous preparation of C₆₀ buckminsterfullerene. The ~10 nm red shift was associated with H₂O dispersions of C₆₀ nanoaggregates (C₆₀)_n that can photosensitize ¹O₂ in their polarizable cores. In contrast to ¹O₂ produced by the water-soluble C₆₀-(γ-cyclodextrin)₂ complex, ¹O₂ trapped inside (C₆₀)_n was short-lived (~2–3 μs), insensitive to solvent and ¹O₂ quenchers, and did not induce photocytotoxicity. To our knowledge, ¹O₂ spectrum from (C₆₀)_n is the most red-shifted ¹O₂ spectrum recorded to date and it may be used to probe the inner polarizability of carbon (nano)aggregates.

Introduction

Buckminsterfullerene C₆₀ is a known photosensitizer of singlet oxygen (¹O₂) in nonpolar solvents (1,2). Much effort has been expended to solubilize C₆₀ in H₂O. One successful strategy is to disperse C₆₀ by different means (3,4) to form nano-aggregates (C₆₀)_n. Solubilization of C₆₀ in aqueous media is driven by biological applications that may involve photosensitization of ¹O₂. The latter is best studied spectrally by its IR phosphorescence. However, we did not observe any correlation between the spectrally determined ¹O₂ production by different aqueous preparations of C₆₀ and their in vitro phototoxicities. Interestingly, we noticed that the spectra of ¹O₂ were often red-shifted in aqueous (C₆₀)_n dispersions which prompted us to examine ¹O₂ production by aggregated C₆₀ in more detail.

Experimental Section

Singlet oxygen lifetimes were measured using a laser pulse spectrometer described in detail elsewhere (6). Briefly, the apparatus utilized a Surelite II laser (Continuum, Santa Clara, CA) for excitation. A germanium diode (Model 403 HS, Applied Detector Corporation, Fresno, California) in conjunction with an efficient optical system was used for signal detection. Data were acquired on a HP 54111D Digitizing Oscilloscope (Hewlett Packard Colorado Springs, CO) interfaced to a PC computer. Singlet oxygen was produced by single pulse excitation of C₆₀ preparations at 355 nm (5 ns duration and ca. 35 mJ energy). The ¹O₂ lifetime was calculated from the monoexponential decay of its phosphorescence observed through an interference filter at 1270 nm.

The steady-state ¹O₂ phosphorescence spectra were recorded on a steady-state ¹O₂ spectrophotometer featuring an optimized optical system as in our pulse ¹O₂ spectrophotometer. Samples were excited from a 500 Watts Hg lamp operating at 300 Watts through a 366 nm interference filter in combination with KG3 heat-removing glass filter. The ¹O₂ phosphorescence spectra were recorded over the range of 1200–1350nm during one ca. 30 s scan and were normalized to the same number of absorbed photons at the excitation wavelength. Perinaphthenone was used as a standard for the ¹O₂ quantum yield calculations.

Absorption spectra were measured using a HP Diode Array Spectrophotometer model 8452A (Hewlett Packard Co., Palo Alto, CA). The relative number of absorbed photons at the excitation wavelength was calculated using the Beer-Lambert law. The samples were prepared in a suprasil fluorescence cuvette (0.5 cm pathlength), and were air-equilibrated or purged with either oxygen or nitrogen gasses, as required.

Transmission electron microscope (TEM) images were taken on a Tecnai-12 Bio-Twin transmission electron microscope (FEI, The Netherlands.) operating at 80 kV. TEM grids were prepared by evaporating approximately 20 μL of aqueous fullerenes solution/dispersion onto a 300 mesh carbon-coated copper grid.

Results and Discussion

The steady-state spectrum of ¹O₂ phosphorescence measured from (C₆₀)_n in O₂-saturated D₂O is shown in Fig. 1, while Fig. 2 shows the corresponding absorption spectrum; the transmission electron microscope (TEM) image of (C₆₀)_n nanoparticles is shown in the insert. Fig.1 also shows ¹O₂ emission spectrum generated by C₆₀ in toluene. The peak position of this spectrum is 1273 nm, while the ¹O₂ spectrum from (C₆₀)_n in D₂O is red-shifted by ca. 10 nm. A similar red shift was consistently observed for (C₆₀)_n dispersed without any additives in H₂O. The red-shifted spectrum in D₂O or H₂O is due to ¹O₂, because it was totally dependent on the concentration of oxygen (5) (Fig.3). Results from time-resolved technique (6) support this conclusion.

Figure 1.

¹O₂ emission spectra from C₆₀ in toluene and (C₆₀)_n in D₂O in the absence and presence of strychnine (toluene) and NaN₃ (D₂O). While NaN₃ at 2 mM did not show any ¹O₂ quenching (not shown), the ¹O₂ quenching by NaN₃ at 100 mM may also involve C₆₀ excited state(s). Spectra in toluene were reduced ca. 4 times; λ_ex=366 nm.

Figure 2.

Absorption spectra of C₆₀ in toluene and C₆₀-(γ-CD)₂ and (C₆₀)_n in D₂O. Insert shows TEM image of (C₆₀)_n.

Figure 3.

¹O₂ emission spectra from (C₆₀)_n in D₂O saturated with oxygen, air and nitrogen.

Surprisingly, the ¹O₂ lifetime (τ_so) photosensitized by the (C₆₀)_n dispersion in D₂O was the same as that in H₂O, even though τ_so is ca. 50 μs in our D₂O sample (7), compared to 3.1 μs in water (8). Fig. 4 shows the decays of ¹O₂ produced by (C₆₀)_n and C₆₀-(γ-CD)₂ in D₂O, and by C₆₀ in toluene. The ¹O₂ transient decay of (C₆₀)_n dispersed in H₂O then extracted into toluene is also presented. The decay of ¹O₂ produced by C₆₀-(γ-CD)₂ confirms that τ_so in our D₂O sample is ~50 μs. A lack of τ_so increase in (C₆₀)_n/D₂O suggests that ¹O₂ emission was not from bulk D₂O, but rather from the (C₆₀)_n core (9). To test this assumption we applied two efficient physical ¹O₂ quenchers: NaN₃ in H₂O/D₂O (10), and strychnine (11) in toluene. We also used a chemical quencher 2,5-dimethylfuran, which was effective in solution but not in the nanoaggregates (not shown).

Figure 4.

Decays of ¹O₂ emission photosensitized by (C₆₀)_n and C₆₀-(γ-CD)₂ in D₂O and by C₆₀ in toluene. Back extraction of (C₆₀)_n from D₂O to toluene (adjacent curve) produced the same τ_so=28.7μs. Decays were produced by single laser pulses 355 nm.

Quenchers provide information on ¹O₂ fate in solution. However, if the quenchers are too large to penetrate the aggregates, they will not quench ¹O₂ there. The cavities in (C₆₀)_n seem to be small and penetrable only by dissolved O₂, which facilitates ¹O₂ production. We found that, indeed, whenever ¹O₂ spectrum was red-shifted, it was largely unaffected by quenchers present at concentrations that totally abolish the usual ¹O₂ emission. These observations support our notion that ¹O₂ must be photoproduced and then deactivated mostly inside the (C₆₀)_n aggregates and not in solution. The τ_so of ¹O₂ inside the (C₆₀)_n is solvent-independent; its estimated value, ca. 3 μs, is close to the τ_so in H₂O. It seems that the conjugated double bond system of C₆₀ quenches ¹O₂, thus controlling its lifetime in the (C₆₀)_n microenvironment. The π electrons are also responsible for the high polarizability of the C₆₀ aggregates (vide infra). These results show that two milieus must be considered for proper analysis of ¹O₂ photosensitization by aggregated C₆₀ in aqueous solutions. We modeled this by using C₆₀ and γ-cyclodextrin (γ-CD).

We prepared non-aggregated water-soluble C₆₀ by complexation with γ-CD (12,13). The ¹O₂ spectrum photogenerated by the C₆₀-(γ-CD)₂ complex in D₂O does not appear to be red shifted and is sensitive to ¹O₂ quenchers (Fig. 5). However, upon total quenching of the emission from D₂O, there was a small contribution from the red-shifted ¹O₂ spectrum, revealing the presence of (C₆₀)_n (Fig. 5). Relative contribution from the (C₆₀)_n emission was even higher for H₂O/γ-CD (Fig.5) due to the shorter τ_so(H₂O) diminishing the usual ¹O₂ emission more in H₂O than in D₂O. C₆₀-(γ-CD)₂ is not stable in H₂O forming yellow colored (C₆₀)_n over time, a process that is accelerated by heat. Thus, both soluble monomeric C₆₀ and dispersed (C₆₀)_n nanoparticles can be present simultaneously in aqueous solutions containing γ-CD.

Figure 5.

¹O₂ emission spectra from C₆₀-(γ-CD)₂ in D₂O and H₂O in the absence and presence of NaN₃ (D₂O). As expected, the spectrum photosensitized by perinaphthenone (P) was completely quenchable by azide; NaN₃ (2 mM) also totally quenched “the usual” ¹O₂ emission from C₆₀-(γ-CD)₂ (not shown). λ_ex=366 nm.

The existence of two clearly different milieus may affect calculation of ¹O₂ quantum yield (Φ_so) based on spectral data. Φ_so is usually measured against a known standard applied in the same solvent, which ensures the same nonradiative (1/τ_so) and radiative (k_r) rate constants for ¹O₂ decay. While both constants are solvent-dependent (14–17), k_r is more difficult to measure and compensation between different environments is more complicated due to lack of reliable values. Obviously, simple comparison of ¹O₂ spectra may yield incorrect Φ_so, if more than a single milieu is involved in ¹O₂ processes. We measured Φ_so by different C₆₀ preparations (Table 1) considering two milieus, when pertinent.

Table 1.

The quantum yield of ¹O₂ photosensitization (Φ_so) by different C₆₀ preparations.

C60 Preparation	Φso	Φsosolution	Phototoxicity
C60/toluene	1	1	N/A
(C60)n/D2O/H2O	~1*	0	No
C60-(γ-CD)2/D2O	1	1	Yes

Measurements used perinaphthenone as a ¹O₂ photosensitization standard (24).

^*For calculations in (C₆₀)_n/D₂O, k_r was estimated from an extrapolation based on the α dependence (16), while 1/τ_so was measured inside (C₆₀)_n and in solutions. The following values were used: k_r, 0.21, τ_so 54 μs in D₂O, and k_r, 3.8, τ_so 2.9 μs in (C₆₀)_n. All solutions were saturated with oxygen, and λ=366 nm was used for excitation. It is obvious that Φ_so calculations for different (C₆₀)_n preparations could produce a range of nonsense values (not shown), if a single milieu would had been considered. N/A, not applicable.

Our measurements for C₆₀ in solution are in general agreement with the published Φ_so values compiled by Prat et al. (18) who also investigated the inclusion complexes of C₆₀ with γ-cyclodextrin. These complexes featured close to unity triplet quantum yields and high Φ_so values (18). For C₆₀ nanoaggregates, the accuracy of our φ estimation in the aggregates is determined primarily by the unknown radiative rate constant. This constant cannot be eliminated (reduced) from the equation by the usual application of ¹O₂ standard, as there is no ¹O₂ reference standard small enough to be placed into the cavity of C₆₀ aggregates. Therefore, our φ estimate (Table 1) is based on extrapolation from the published polarizability dependence (16). While this arbitrary procedure can produce φ values larger than unity, it does suggest that φ is close to unity in oxygen purged solutions when the aggregate cavities are saturated with oxygen molecules.

Notice that physicochemically correct Φ_so values may not always correlate with ¹O₂-induced photocytotoxicity. A simple correlation is expected only for ¹O₂ production in solution, but not inside the (C₆₀)_n particles that deactivate ¹O₂ prior to diffusion out. Thus, only C₆₀-(γ-CD)₂, which is a perfect ¹O₂ generator (Table 1), was phototoxic to cell cultures, while (C₆₀)_n was not (to be published later). A correct understanding of ¹O₂ generation by aqueous C₆₀ preparations allows us to reliably predict phototoxicity by considering ¹O₂ sensitivity to solvent and quenchers, which concurs with the spectral shift of ¹O₂ emission.

What is the origin of this red shift? Unlike k_r and τ_so, ¹O₂ emission energy is less affected by solvents. Nonetheless, red shifts have been observed (19–21) and linked to increasing solvent polarizability (α) (16). The most red-shifted ¹O₂ spectra were in iodinated solvents (16), reaching a maximum at around 7824 cm⁻¹ (1278 nm). The ¹O₂ spectrum that we measured in methylene iodide (n=1.74) shows similarly red-shifted (~9 nm) peak position (not shown). The spectrum in (C₆₀)_n is even more red-shifted (7798 cm⁻¹), extrapolating (16) to α.≈0.46 that would match solvent refractive index n=1.887 (in diamond n=2.417 (22)). Moreover, the red shifts were empirically associated with the increasing k_r values (16), hence, k_r should also be enhanced in (C₆₀)_n. We used this association to estimate k_r inside (C₆₀)_n for ¹O₂ quantum yield calculation shown in Table 1.

The above α estimate reveals a very polarizable (C₆₀)_n core probed by ¹O₂. This core is hydrophobic and not accessible to solvent but to O₂ that slowly diffuses in and out with a dwelling time longer than 3 μs (¹O₂ cannot diffuse out in its lifetime). While at least four C₆₀ must aggregate to create a protective cavity for O₂, the average (C₆₀)_n size in our preparation estimated from the TEM images was n=30–60 molecules. (C₆₀)_n seems to be a dynamic formation showing Brownian motion that creates cavities large enough (23) to host O₂ that, upon excitation, can convert to ¹O₂ (Fig. 5).

In summary, to our knowledge, we found the most red-shifted ¹O₂ phosphorescence spectrum recorded to date generated in aqueous preparations of (C₆₀)_n. The ¹O₂ spectral shift concurs with ¹O₂ sensitivity to solvent and quenchers, which may be used to reliably predict phototoxicity exerted by ¹O₂ in different preparations of carbon nanoaggregates. Moreover, since the ¹O₂ spectral red shift correlates well with polarizability, ¹O₂ acts as a probe providing a unique insight into the inner polarizability of (C₆₀)_n aggregates.