Surface Photochemistry: Benzophenone as a Probe for the Study of Modified Cellulose Fibres

1 Centro de Quı́mica-Fı́sica Molecular Complexo Interdisciplinar, Instituto Superior Técnico, Technical University of Lisbon, Av. Rovisco Pais, 1049-001 Lisboa, Portugal 2 Laboratório de Quı́mica Orgânica, Departamento de Engenharia Quı́mica, Instituto Superior de Engenharia de Lisboa, R. Conselheiro Emı́dio Navarro 1, 1950-062 Lisboa, Portugal 3 Escola Superior de Tecnologia e Gestão, Instituto Politécnico de Portalegre, Lugar da Abadessa, APT 148, 7301-901 Portalegre, Portugal 4 Laboratoire Sciences des Matériaux et Environnement (LMSE), Faculté des Sciences de Sfax, BP 802-3018 Sfax, Tunisia 5 Laboratoire ITODYS, Université Paris Diderot, CNRS, 75005 Paris, France

Benzophenone (BZP) is an extremely useful molecule for probing new hosts. The n → π * absorption transition was found to be very sensitive to the environment characteristics and also exhibits a photochemistry which depends on the host properties [10,12,14]. In a recent paper [14], we reported a comparative study of the luminescent properties of BZP adsorbed onto reversed phase silicas, "normal" silica, and silicalite (a de-aluminated zeolite). Apart from the triplet-state luminescence observed in all cases, in the case of "normal" silica the emission of an excited form of hydrogen-bonded benzophenone was also detected [14].

Research Letters in Physical Chemistry
With this work we intend to contribute to the study of the mechanism by which modified cellulose fibres are able to trap dissolved organic pollutants from water. Indeed, as we have shown previously [15,16], grafting of linear alkyl chains on the fibre's surface boosts its capacity to uptake organic solutes from aqueous solution. By ensuring grafted alkyl chains, one gives rise to hydrophobic domains on which organic solutes can be accumulated. The adsorption process occurs by transfer of the sparingly soluble organic molecules from water to organic zones of the modified cellulose where the most significative adsorbent-adsorbate interactions occur.
This work reports the use of BZP, a very well characterized probe, to study new hosts (i.e., modified celluloses). The modification consists in grafting with alkyl chains, bearing 12 carbon atoms, by surface esterification with a high density of alkyl chains [15], therefore, transforming the polar surface of the normal cellulose into surfaces with a certain degree of nonpolar character. A comparison of the photochemical behaviour of BZP in modified and nonmodified celluloses was made.

Materials
Microcrystalline cellulose (Fluka DS0) was used as powdered solid support, as received. Benzophenone (Koch-Light, Scintillation grade) and ethanol (Merck, LiChrosolv grade) were also used as received. The preparation of the modified cellulose fibres started with the use of microcrystalline cellulose and involved an acylation reaction based on a solvent exchange procedure, as described in detail in [15]. The aliphatic anhydrides have 12 carbon atoms (C12) per alkyl chain and the final modified cellulose has 1500 (C12-1500) or 1700 (C12-1700) micromoles of alkyl chains per gram of cellulose [15].

Sample preparation
Benzophenone adsorption on samples was performed using two methods: the solvent evaporation method for the case of ethanol [14], and, also, adsorption from water [15]. The former method consists in the addition of an ethanolic solution containing the probe to the previously dried powdered solid substrate, followed by solvent evaporation from the stirred slurry in a fume cupboard. In the case of water, the fibers were first swollen for at least two hours, and the addition of BZP was done by adding 500 micromoles of this probe dissolved in ethanol (saturated solution so that the added amount of ethanol was minimized). The water suspensions were kept under agitation for 24 hours and the modified cellulose (with the adsorbed BZP) was removed by filtration. From the initial 500 μmole g −1 of BZP, about 300 μmole g −1 remained into the powdered substrate (i.e., an increase in the retention capacity of the powdered substrate of about twelve times when compared to previous reported results) was obtained with nonmodified cellulose [15].
In both cases, the final solvent removal was performed for about two hours in an acrylic chamber with an electri-cally heated shelf (Heto, Model FD 1.0-110) with temperature control (30 ± 1 • C) and under moderate vacuum at a pressure of ca. 10 −3 Torr.

Ground-state diffuse reflectance absorption spectra (GSDR)
Ground-state absorption spectra for the solid samples were recorded using an OLIS 14 spectrophotometer with a diffuse reflectance attachment. Further details are given in [1,7].

Laser-induced luminescence (LIL) and diffuse reflectance laser flash photolysis (DRLFP) systems
Schematic diagrams of the LIL and of the DRLFP systems are presented in [1].

RESULTS AND DISCUSSION
In both cases (benzophenone adsorption from water or from ethanol), similar results were obtained for the two modified celluloses C12-1500 or C12-1700 (within experimental error) (i.e., with degree of modification ranging in 1500-1700 μmol of alkyl chains per gram of cellulose). This is valid for all the experimental techniques used in this work. This paper reports data obtained for this specific degree of alkylation and compares samples obtained with two different solvents.
For BZP/C12-1700 samples prepared by solvent evaporation method with ethanol, the ketone's ground-state absorption S 0 → S 1 transition (n → π * ) has a maximum at about 347 nm and exhibits a clear vibronic structure characteristic of the excited carbonyl group of BZP in a hydrophobic environment [17].
For BZP adsorbed from water onto C12-1500 modified cellulose, the ground-state absorption curves appear now as broad bands, shifted hypsochromically when compared with BZP/C12-1700 ethanol case, where the vibronic absorption bands of the carbonyl group can be seen ( Figure 1). These hypsochromic shifts are quite characteristic of the n → π * transition with increasing polarity of the surface [7,11,14,17]. The broadening of the spectra is probably also related both with heterogeneity of the adsorbent and a much smaller rigidity of the adsorbed probe. Solution absorption spectra of BZP, for instance in cyclohexane and ethanol, also exhibit this type of influence of polarity, characteristic of the n → π * transition [17].
Clearly, going from C12-1700/EtOH to C12-1500/H 2 O an increase in the surface polarity is observed, quite consistent with the surface characteristics and different adsorption sites to BZP: long alkyl chains with 12 carbons in the C12-1700/EtOH sample, and contact with the hydroxyl groups of the cellulose polymer chains in the C12-1500/H 2 O case.  Those time-resolved spectra were obtained with airequilibrated conditions and were identical to the ones obtained with argon-purged samples within experimental error. Half-livesof about 20 μs can be obtained from timeresolved spectra shown in Figures 2 and 3.

Room temperature laser-induced phosphorescence
For comparison purposes, lifetimes of about 80 microseconds were determined for the calixarene inclusion [10,11] and 3.1 milliseconds for inclusion into the narrower channels of silicalite [12,18], as compared to about 40 microseconds for benzophenone microcrystals,all determined at the maximum emission wavelength (about 448 nm) [14].
A new tool for a lifetime distributions analysis of emissions of probes adsorbed onto heterogeneous surfaces was recently developed by our research group [18]. This new   methodology allows for asymmetric distributions and uses pseudo-Voigt profiles (Gaussian-Lorentzian product) instead of pure Gaussian or Lorentzian distributions. A very simple and widely available tool for fitting has been used, the Microsoft Excel Solver. This is a convenient way to treat the emission decay data because it reflects the multiplicity of adsorption sites available for the probe onto each specific surface. The use of a sum of a few exponentials to analyse the decay of probes onto heterogeneous surfaces is a description without physical meaning [18].
A detailed study of the luminescence decay curves of pyrene included within p-tert-butylcalix [4]arene cavities and benzophenone into silicalite channels has been reported recently [18,19].
When applied to the C12-1700/EtOH and C12-1500/H 2 O cases (see Figures 4(a) and 4(b)), and after recording the decay traces in various instrumental time scales (experimental information was obtained starting in the 0.02 microsecond or longer timescales), the lifetimes distribution analysis evidences a single broad band centred at ca. 13 microseconds, which we assigned to BZP adsorbed onto the long alkyl chains with 12 carbons in the C12-1700/EtOH case; but for the case of C12-1500/H 2 O, another band (in this case a sharp band) centred at 2.6 microseconds predominates, which we assigned to BZP in close contact with the hydroxyl groups of the cellulose polymer chains. The emission from BZP in contact with the alkyl chain still exits (peaking now at 19 microseconds) but it is of much less importance than the other component.
This kinetic information is consistent with previous data from ground-state absorption spectra as well as the spectroscopic information from laser-induced luminescence experiments.
The laser-induced emission experiments for BZP/C12-1500/H 2 O also showed a special emission of BZP in the nanosecond time scale, very similar to the one reported in [20, Figure 3(b)] for BZP on MCM-41. This emission peaks at ca. 430 nm, and therefore it originates from hydrogenbonded BZP (data not shown). In the case of the BZP/C12-1700/EtOH sample, no hydrogen-bonded BZP emission in the nanosecond time scale could be detected.
Previous work indicates that calixarene [10][11][12], cellulose [7,8], or reversed phase silicas [14] are good hydrogen atom donors towards BZP (in solid powdered samples), so it sounds reasonable to assume that ketyl radical formation of benzophenone may also occur here. The emission maxima at 575 nm presented both in spectra of Figures 2 and 3 are a reasonable indication that this is also the case.
In order to perform the lifetime distribution analysis, several decay curves in different instrumental time scales were recorded. A superposition of those decay traces was made by normalization of each decay curve at a time range where they overlap, in order to produce a composite decay with closely spaced data at short times and larger spaced values at long times. This procedure was adopted before [14,19], and is necessary because the abscissa is ln t, therefore a very large time range had to be used.

Diffuse reflectance laser flash photolysis
Time-resolved absorption spectra of samples of BZP/C12-1700/EtOH and C12-1500/H 2 O samples were obtained by the use of diffuse reflectance laser flash photolysis technique, developed by Wilkinson et al. [2][3][4]. In this study, the use of an intensified charge-coupled device as a detector allowed us to obtain time-resolved absorption spectra with nanometer spectral spacing (i.e., where the 200-900 scale is defined by the 512 pixels used for recording spectra in the array of the ICCD) [1,[8][9][10][11][12][13][14].
Both transient absorption spectra of the BZP/C12-1700/EtOH and C12-1500/H 2 O samples have shown the simultaneous formation of triplet benzophenone and also of hydroxylbenzophenone radical (BZP • OH) (data not shown). The triplet-triplet absorption spectra of benzophenone (max. at 530 nm) was easily identified from comparison with the one published by Wilkinson et al. [4]. The transient absorption which peaks at 390 nm can be assigned to BZP • OH radical by comparison with previously reported spectra in solution or on the MCM-41 surface [20]. Ketyl radical formation peaking at about 550 nm was observed for both BZP/C12-1700/EtOH and C12-1500/H 2 O samples a few microsecond after laser pulse, showing that hydrogen atom abstraction can occur either when adsorbed onto the long alkyl chains or when BZP is close to the main polymer chain of cellulose where the hydrogen atoms, bound to the secondary carbons, are the most easily abstractable ones [7].

CONCLUSIONS
The photochemistry of BZP onto the modified celluloses is determined by the nature of the adsorption site (long alkyl chains or the hydroxyl groups of the polymer chains). The adsorption site of the probe depends on the solvent used for sample preparation: ethanol privileges the first case while water leads to the second situation. As a consequence of the detected BZP phosphorescence, BZP ketyl radical fluorescence and hydrogen-bonded BZP luminescence reflect the different sites for adsorption.
A lifetime distributions analysis provided important information and revealed an important quenching effect in the case of adsorption from water, comparable to the case of adsorption onto microcrystalline cellulose [7], due to the more efficient hydrogen abstraction reaction from the glycoside rings of cellulose when compared with hydrogen abstraction from the alkyl chains of the modified celluloses.
Diffuse reflectance transient absorption spectra revealed the presence of the triplet state of BZP in all supports under study, and also of the diphenylketyl radical and BZP • OH radicals.