Spectroscopic Characterization of Neutral and Cation Radicals of α

Sep 16, 2010 - Department of Physics, Norwegian University of Science and Technology .... The Journal of Physical Chemistry A 2011 115 (29), 8242-8247...
0 downloads 0 Views 1MB Size
Download PDF
J. Phys. Chem. A 2010, 114, 10795–10802

10795

Spectroscopic Characterization of Neutral and Cation Radicals of r-Tocopherol and Related Molecules: A Satisfactory Denouement K. Razi Naqvi,*,† Heng Li,† T. B. Melø,† and Richard D. Webster‡ Department of Physics, Norwegian UniVersity of Science and Technology (NTNU), NO-7491 Trondheim, Norway, and DiVision of Chemistry and Biological Chemistry, School of Physical and Mathematical Sciences, Nanyang Technological UniVersity, Singapore 637371 ReceiVed: July 20, 2010; ReVised Manuscript ReceiVed: August 18, 2010

Notwithstanding the facile occurrence of one-electron oxidation in R-tocopherol and its acetate (TOH and TOAc, respectively), and despite the remarkable stability, under appropriate conditions, of the oxidation products (TOH•+, TO•, and TOAc•+), their spectroscopic characterization is in an unsatisfactory state, calling for a fresh attempt to acquire reliable data. A new, model-free method is developed for analyzing time-resolved spectra showing the progress of the reaction TOH + R• f TO• + RH, where R• is a stable free radical. The resulting absorption coefficients of TO• in dichloromethane and hexane are in severe disagreement with some recent values derived from stopped-flow spectrophotometry. The discrepancy is traced to the imposition of boundary conditions that do not take proper account of the dead time of the apparatus; when multiplied by a factor of two, the stopped-flow data fall mostly in the range ε ) (7.5 ( 0.5) × 103 M-1 cm-1, conforming with the results of this study and the values found by Boguth and Niemann in 1969. Absorption spectra of the radical cations produced (electro)chemically are found to be reliable only in the visible region. Incomplete conversion of the parent compound to the radical cation, an obstacle to the determination of absorption coefficients from electrochemical studies, is circumvented by combining EPR and optical spectroscopy. The absorption coefficients of TOH•+ and TOAc•+, determined in this manner, are found to be, respectively, 1.6 × 104 and 1.3 × 104 M-1 cm-1, in accord with the values found first through similar means. Introduction Amongst the phenolic antioxidants, R-tocopherol (TOH) has been recognized as the most potent representative.1,2 Its immense importance in biology and its attraction for researchers from diverse disciplines stem from several factors, among them its membership of the vitamin E family, its role in natural systems, where it is believed to guard against lipid autoxidation and to provide photoprotection, and the extraordinarily long life of the tocopheroxyl radical TO•,3 and even that of the radical cation TOH•+ under acidic conditions.4 It has also been suggested that TOH plays yet another role, independent of its antioxidant properties, by inhibiting vascular smooth muscle cell proliferation.5 While analyzing the results obtained recently in an ongoing investigation of the photochemical behavior of TOH, R-tocopherol acetate (TOAc), and other related molecules,6,7 we needed to know the absolute absorption spectra of the following species, all of which are paramagnetic: TOH•+, TO•, and TOAc•+. Since spectroscopic characterization of a long-lived radical is not regarded as a particularly demanding task, and the species of our interest have been the subject of numerous investigations, it seemed that the required information could be retrieved simply by scouring the literature. In the brief survey presented below, pump-probe investigations have been excluded, since this approach to finding the absorption coefficients of the aforementioned species will be the subject of a separate paper. * To whom correspondence should be addressed. E-mail: razi.naqvi@ ntnu.no. † Norwegian University of Science and Technology. ‡ Nanyang Technological University.

It is now 40 years since Boguth and Niemann8 generated TO• chemically, and recorded its absorption spectrum with the aid of a rapid scanning spectrometer; a kinetic analysis of the rate • ) 6.7 × 103 of decay of TO• led them to conclude that ε423 • M-1 cm-1 in benzene and ε426 ) 8.1 × 103 M-1 cm-1 in chloroform, where ε• denotes the molar absorption coefficient, and the subscript specifies p ≡ λ1/nm, where λ1 is the wavelength of the first peak. Although the shape of the spectrum has been confirmed by later investigators, who too used chemical oxidation, their values of ε•p range from 2.5 × 103 M-1 cm-1 in hexane to around 4 × 103 M-1 cm-1 in chloroform, dichloromethane (DCM), ethanol, and toluene.9-11 The need for a reevaluation of ε•p for the tocopheroxyl radical, and for extending the spectrum to shorter wavelengths, where ground-state molecules also absorb, can hardly be overstated. In 1974, Svanholm and coauthors12 reported that oxidation of MOH (a model compound in which the phytyl chain of TOH is replaced by a methyl group) in trifluoroacetic acid (TFA) or in TFA-DCM results in the reversible formation of MOH•+. They stated that the visible absorption spectrum of the “yellow green” solution of the oxidation product of MOH “in TFA showed a maximum at 456 nm (ε 610)”; in our notation, we xM ) 6.10 × 102 M-1 cm-1. More will express their result as ε456 than a decade later, the visible absorption spectra of TOH•+ and TOAc•+ were recorded in TFA and were found to be rather similar in shape: asymmetric bands peaking at about 465 nm and skewed toward shorter wavelengths.4,13 The following values were reported for the molar absorption coefficients: εx465 ) 1.82 xAc ) 1.26 × 104 M-1 cm-1 × 104 M-1 cm-1 for TOH and ε465 for TOAc. More recently, UV-visible spectra recorded during the electrochemical oxidation of the tocopherols (and two related model compounds, including MOH) in an organic solvent, with

10.1021/jp106736x  2010 American Chemical Society Published on Web 09/16/2010

10796

J. Phys. Chem. A, Vol. 114, No. 40, 2010

a supporting electrolyte and an organic acid (for suppressing the deprotonation of the cation), have become available;14-16 the choice of the solvent, acetonitrile or DCM, did not exert any noticeable influence on the spectra, and the shapes of the visible bands were in fair agreement with those observed in TFA.4,13 On the other hand, a recent visible absorption spectrum of TOAc•+, with p ) 492 in DCM,17 bears hardly any resemblance to the other spectra mentioned above. It will be x shown later that the electrochemical spectra15,16 imply that ε465 3 -1 -1 is close to 5 × 10 M cm . One is left with the impression that the results obtained heretofore are so variable, uncertain, and inconclusive that their utility may well be called into question. The traps and pitfalls that await an investigator of this problem include, we feel, the presence, on the one hand, of unsuspected impurities in the solvent and the imposition, on the other hand, of inappropriate boundary conditions on the kinetic rate equations. The purpose of this paper is to present the results of a new investigation, where a model-free global analysis of time-resolved spectra is used for finding the absorption coefficients, and to reconcile the calculations of the authors of refs 10 and 11 with those of Boguth and Nieman, whose results have been corroborated by our analysis. Experimental Methods A commercial scanning spectrophotometer (Shimadzu UV160A) was employed for recording steady-state absorption spectra. Time-resolved absorption spectra were acquired by using one of the two home-built fiber-optic multichannel spectrometers, or a commercial photodiode array (PDA) spectrophotometer (Agilent 8453). The last named instrument was used for the sole purpose of duplicating an experiment performed by Nakanishi and coauthors,18 who used such an instrument to record a set of spectra at intervals of 2 s. When spectra recorded automatically at regular intervals were required, we used, in all other cases, a CCD spectrometer; its main components are a 2048 channel spectrograph (B&WTEK Model BRC642E), which covers a wide spectral range (190-1050 nm), and a light source (Hamamatsu L10290) that emits in the 200-1100 nm region. For recording spectra manually at irregular intervals, an older PDA spectrometer was employed;6 its disperser-detector component is a Zeiss 1024 channel spectrograph (Zeiss MCS 224) covering the 200-1010 nm range, and the source of monitoring light is a miniature xenon flash lamp. The dead time of the CCD-based arrangement was determined mainly by the time needed for mixing and is estimated to be 2-3 s, and that of the PDA-based instrument (about 10 s) by the time needed to execute a manual command for saving a data stream. The spectra shown here were obtained by using the former instrument; although slower and far less sensitive than the CCD spectrometer, the latter instrument, which has a superior signal-to-noise ratio, was used primarily for confirming the results furnished by the CCD spectrometer. Chemical oxidation for generating TO• was carried out as follows. High-purity argon (g99.999%) was bubbled for 15 min through an absorption cuvette containing 2 mL of a solution (20-80 µM) of 1,1-diphenyl-2-picrylhydrazil (dpph•) before recording its absorption spectrum (an average of 500 spectra), and a gentle flow of the gas was maintained throughout the experiment. With the CCD spectrograph, acquisition of a series of spectra at regular intervals (usually 0.5 or 0.8 s) was started before an aliquot (20-50 µL) of a TOH solution was squirted into the cuvette as soon after the start as possible. Data collection was stopped after the acquisition of 800 intermediate spectra,

Naqvi et al. after which the final spectrum was recorded (again as an average of 500 spectra). When the PDA spectrograph was used, only one spectrum could be stored after each command; in this case, the spectra were recorded at irregular intervals and the total number of spectra was much smaller (typically 10). The radical cations TOH•+ and TOAc•+ were produced either by dissolving the solute in TFA or by adding small amounts (20-50 µL) of TFA to a solution (in DCM) of the parent compound. The sample was transferred to the EPR spectrometer when its absorbance had reached a stationary value. After the completion of the EPR measurement, the absorbance of the sample was measured again to verify that the concentration of the absorbing species had not changed meanwhile. The EPR spectra were obtained with a so-called free radical monitor (JEOL JES-FR30). The sample, contained in a flat quartz cell (0.5 mm thick walls with a separation of 0.2 mm) with a long capillary neck, was placed in the middle of the microwave cavity. The operating conditions for the EPR spectrometer were as follows: power, 4 mW; modulation width, 0.05 mT; center of magnetic field, 335 mT; sweep time, 1 min; sweep width, 5 mT; time constant, 0.1s; amplification, 20. The solvent used for the reference sample was acetonitrile, which gave a resonance similar to that seen in the TFA solutions of TOH and TOAc; as an additional check on our measurements of the relative signal intensity, we used the ratio between peakto-peak amplitudes of the signals from the sample (or the reference) and a Mn2+ marker attached to the cavity of the spectrometer. In electrochemical studies, in situ UV-vis spectra were recorded with the aid of a scanning spectrophotometer (PerkinElmer Lambda 750); the solution was placed in a semitransparent electrochemical cell (with a Pt mesh working electrode) whose temperature could be varied from ambient to 233 K. The spectral reconstruction method, used for global analysis of the spectra of mixtures, and the rationale of the method have been detailed in previous publications.19,20 Singular value decomposition (SVD) analysis of a set of electrochemical spectra was performed by using the Matlab 7.8 software package (The Math Works Inc., Natick, MA). All chemicals were purchased from Sigma; the solvents were all of spectroscopic grade, but anhydrous acetonitrile and DCM were also used and were found to yield more consistent results; only those data which relate to supposedly low-moisture samples are reported here. The concentrations of the solutes were calculated by using the values of molar absorption coefficients determined in our laboratory. Unless otherwise stated, the spectra reported below were recorded at 295 K. Results and Analysis Because our experiments will provide values of the three •d absorption coefficient (ε•p, εxp , and εxAc p ) relative to εp , the • absorption coefficient of dpph , we measured the molar absorption coefficients of dpph•, TOH, MOH, and TOAc; these data are presented in Table 1; each entry is in the form µ ( σ, where µ and σ are the mean value and the standard error of five measurements based on separate weighings. The rest of this section will explain the steps that have led us to the absorption coefficients of the oxidation products, which have also been included in Table 1. Chemical Oxidation. Reaction with dpph•. Although we will examine the reaction between TOH and dpph•, our comments are sufficiently general and are therefore applicable to similar systems involving other phenols and stable free radicals different from dpph•.10,21,22 We will continue to focus attention on TOH,

Neutral and Cation Radicals of R-Tocopherol

J. Phys. Chem. A, Vol. 114, No. 40, 2010 10797

TABLE 1: Spectroscopic Data species

solvent

T (K)

TOH TOAc dpph• dpph• dpph• TO• TO• MO• TOH•+ TOH•+ TOAc•+

hexane hexane hexane dichloromethane acetonitrile hexane dichloromethane acetonitrile trifluoroacetic acid dichloromethane trifluoroacetic acid

295 295 295 295 295 295 295 295 295 295 233

10-3 × ε (M-1 cm-1) 3.40 ( 0.03 2.03 ( 0.04 10.6 ( 0.1 11.5 ( 0.4 11.9 ( 0.1 6.5 ( 0.5 7.1 ( 0.4 6.0 ( 0.6 15.8 ( 0.5 12.9 ( 1.0

λ1 (nm) 298 288 509 528 527 418 426 426 461 465 464

but we will allow for other free radicals and use the more general symbol frad•. Mixing of TOH and frad• leads to the reaction depicted below:8 k1

frad• + TOH a TO• + fradH k-1

(1)

The tocopheroxyl radical TO• may also react with frad•, or with itself:21,22 k2

TO• + frad• 98 tocopherol oxidation product + fradH

(2) k3

TO• + TO• 98 disproportionation or dimerization

(3) Since we will be facing the need to specify the concentrations of the species of interest at various instants, it will not be convenient to adhere throughout to standard chemical notation (of using square brackets for denoting the concentration of the enclosed species) and the customary mathematical notation (of displaying the argument of a function within parentheses). Accordingly, we introduce the following abbreviations: E ≡ [TOH], F ≡ [frad•], R ≡ [fradH], and N ≡ [TO•]. For the reader’s convenience, we will occasionally state a mathematical expression using both alternatives, and it may also be helpful to note that R denotes the concentration of the reduced form fradH, and N represents the concentration of the neutral tocopheroxyl radical. The relevant rate equations can now be expressed as as

dF ) -k1EF + (k-1NR - k2FN) dt

(4a)

dN ) k1EF - (k-1NR + k2FN + k3N2) dt

(4b)

The premixing values of E and F will be denoted as E0 and F0, respectively; and that of N will be kept at zero by experimental design. The initial conditions to be imposed on the above rate equations will not be discussed at this stage. The condition E0 . F0, frequently enforced for simplifying the determination of the rate constants (k1, k2) and/or ε•p,9-11 is not

necessary, and not even desirable, in our experiment, where the dead time is larger than 1 s. The above reaction scheme assumes ideal conditions. Now, Musialik and Litwinienko23 as well as Foti and coauthors24 have emphasized that the apparent value of k1 is strongly influenced by the solvent impurities that can affect the ionization equilibrium of TOH. Since the value of k1 is not needed here, such impurities will cause no problem, provided that the available time resolution is adequate for following the course of the reaction. Any impurities capable of reacting with frad• in competition with TOH (when its concentration is relatively low) can be taken into account by replacing k1E ≡ k1[TOH] on the right-hand side of eq 4a by (k0 + k1E), with the understanding that k0 ≡ ΣmkmIm, where Im denotes the concentration of a reactive impurity and km is the corresponding rate constant. We have also assumed that, when the progress of the reaction is monitored through absorption spectrophotometry, the monitoring light does not bleach frad•, which is a legitimate concern raised by Brault and coauthors.21,22 Likewise, we have ignored the formation of TO• as a result of photoexcitation of TOH by the monitoring light. The validity of these assumptions will be examined later. Recently, Mukai and coauthors10,11 conducted an extremely thorough investigation of the growth and decay of TO•, using aryloxyl radical (ArO•) instead of dpph• and a stopped-flow spectrophotometer for monitoring the concentration of TO• in several solvents, which will be classified here as compliant or noncompliant. They determined ε•p by fitting their data to a kinetic model that can be recovered from eqs 4a-b by setting k-1 ) 0 ) k2. A solvent will be called compliant if the decay of N conforms, after an initial build-up, to second-order decay; otherwise, it will be called noncompliant. Benzene, chloroform, DCM, ethanol, and toluene fall in the former category, whereas diethyl ether, heptane, and hexane were found to be noncompliant. Others8,9 have used a simpler approach. We will return to a discussion of previous kinetic analyses after describing our own strategy and presenting the results to which it has led us. We are going to argue that, in the initial phase of the reaction, the terms within the parentheses in eqs 4 may be ignored, and one may write dN/dt ) -dF/dt; this implies that, if τ is a sufficiently small interval, the diminution in F during this time interval, namely F0 - F(τ), will equal N(τ) - N(0) ) N(τ). • Let A•d p denote the absorbance lost by dpph and Ap the • absorbance gained by TO during a given time interval τ; a plot of x ≡ A•p(τ) against y ≡ A•d p (τ) for different values of τ will follow, according to the above reasoning, a rectilinear trend, with a slope b that can be equated to ε•p/ε•d p ; we will refer to a plot of this kind as a calibration plot. When the term k2FN ≡ k2[frad•][TO•] becomes non-negligible, some dpph• will be lost through reaction 2, and the gain in the concentration of TO• will be reduced by the intervention of reaction 3; if we also take into account the term k3N2 ≡ k3[TO•]2, and the neglect of k0 in eq 4a, we are led to conclude that our method can only provide a lower limit for the ratio ε•p/ε•d p . Since DCM, a compliant solvent, has proved to be convenient also for electrochemical investigations of TOH•+ and TOAc•+, we begin by discussing the results obtained with this solvent. Figure 1 shows plots of A0(λ) and A4(λ), respectively, where A0, the initial spectrum, represents only dpph•, and A4 is the fourth member of a set of intermediate spectra recorded after the addition of TOH. The last member of the set, A∞, which will not be called an intermediate spectrum, was recorded after the disappearance of dpph• as well as TO•. At wavelengths longer than 310 nm, where TOH does not absorb, one may

10798

J. Phys. Chem. A, Vol. 114, No. 40, 2010

Naqvi et al.

Figure 2. Absorption spectrum of TO• in hexane (solid curve); the other two curves represent the consequences of a 40% error in subtracting the absorbance of TOH.

Figure 1. A0 is the absorption spectrum of a 26 µM solution (2 mL in volume) of dpph• in DCM at 295 K. A4 is the fourth of a series of intermediate spectra (A1, A2, ... An) recorded (0.8 s apart) after a solution of TOH in DCM (40 µL, 3 mM) was added to the cuvette, and A∞ is the final spectrum. Σ4 is a linear combination of A0 and A∞; for λ g 460 nm (the region to the right of the dashed vertical line), Σ4 is indistinguishable from A4, and the difference A4• ) A4 - Σ4 is the spectrum of TO•; the upper inset shows normalized plots of Ai• (points) and the average (line) of six spectra (A3•, A4•, ... A8•). In the lower inset, x (y) is the decrease (increase) in the absorbance of dpph• at 528 nm (TO• at 426 nm).

regard the sample as a three-component mixture (one reactant and two products); at wavelengths longer than 470 nm, where even TO• does not absorb, one need consider only two components. Accordingly, Ai can be reconstructed, for λ > 470 nm, as a linear combination of the initial and final spectra; that is to say, we can write

∑ (λ) ) RiA0(λ) + βiA∞(λ)

(5)

i

in which Ri and βi stand for optimized coefficients found through linear regression. If a fit Σi is extended to wavelengths significantly shorter than 470 nm, it will no longer coincide with Ai, and the difference would represent, so long as one does not go to wavelengths shorter than 310 nm, the absorbance contributed by TO•. Inspection of Figure 1 confirms that, for λ g 460 nm, a plot of Σ4(λ) is indeed indistinguishable from A4(λ). The absorption spectrum of TO• can now be found, for λ g 310 nm, by forming the difference between A4 and the spectrum obtained by extending Σ4(λ) to the 310-460 nm region (the solid curve identified by the arrow). The shape of A•i ) Ai - Σi was almost independent of i, at least for the first 10 spectra; the upper inset compares the shape of A•i (i ) 3, 4, · · · 8), which have all been normalized at 426 nm, with their average. A calibration plot for this data set is shown in the lower inset. The initial linear behavior is consistent with our prediction and provides a reliable value (0.535 ( 0.004) of the slope; for larger values of τ, the experimental points (not shown here) fall below the straight line, showing the growing importance of the term involving N2 in the right-hand side of eq 4b. A different run led to a calibration plot with b ) 0.562 ( 0.004; the uncertainty in the value of b for a single run will no longer be quoted because it is very much smaller than the difference in the values of b in two different runs. Because the sublinear portion of the calibration plot becomes significant only when τ exceeds 10 s, we also used the PDAbased spectrometer; to compensate, at least partially, for the

longer dead time of this instrument, the slope (now denoted by b1) was calculated by using only A1, the first intermediate spectrum; in other words, b1 ≡ y1/x1 was equated to ε•p/ε•d p . Two runs provided the values b1 ) 0.61 and 0.63. Our discussion of the consequences of departure from nonideal experimental conditions makes it patent that the largest value of b (or b1) obtained in a laboratory is to be preferred, other things being equal, of course; accordingly, we will take • ε•p/ε•d p ) 0.62 ( 0.1, which leads to the result ε426 ) (7.1 ( 0.4) 4 × 10 M cm. • The above value of ε426 in DCM is in severe disagreement • ) 4.1 × 103 M-1 cm-1 was with ref 10, where the result ε427 stated for DCM solutions. For hexane, a noncompliant solvent, • ) 2.5 × 103 M-1 cm-1. an even smaller value was found: ε418 We therefore decided to use our method for finding the absorption coefficient of TO• in hexane solutions. The results obtained by mixing TOH and dpph• in hexane are summarized in Table 1 and Figure 2. For plotting the solid curve, we have subtracted the estimated contribution made by the residual absorbance of TOH, using, as a guide, some laser photolysis data (to be published); the dashed (dotted) spectrum was calculated by changing the ground state contribution to 1 ( 0.4 times the value used for the solid curve. The region below 280 nm has not been shown because the shape of the spectrum in this region has not yet been confirmed by a reliable independent measurement. To perform still another check, we decided to investigate the reaction of dpph• with MOH in MeCN, a system examined by Nakanishi and coauthors;18 for this purpose, we used the same type of spectrometer as that employed by these authors. Since our plots (not shown here) are similar to theirs, it is sufficient to state that our calibration plot led to the value of the absorption coefficient of MO• given in Table 1. We close this section by commenting on an important observation, made by Brault and his co-workers,21,22 concerning the difference between the single-wavelength mode for recording a kinetic trace and the so-called reversed-optics mode used in multichannel spectrometers. In the former case, one may (and usually does) interpose a monochromator between a white light source and the cuvette holder, thereby minimizing photodegradation of the sample, which might become a serious problem in the latter case, which requires placing the sample between the white light source and the spectrograph. That all three instruments employed in this work yielded similar results, despite the disparate intensities of their monitoring beams, indicates that the use of reversed-optics spectrometers did not vitiate our data. Reaction with TFA. Gradual addition of TFA to a solution of TOH in DCM gives rise to a series of spectra that show the

Neutral and Cation Radicals of R-Tocopherol

Figure 3. Spectral changes recorded by gradual addition of TFA to a solution (230 µM) of TOH in DCM; all spectra have been normalized at the 461 nm peak. Curve 9, taken from ref 14, is the spectrum of TOH•+ in MeCN (with 0.05 M CF3SO3H and 0.5 M Bu4NPF6).

build-up of TOH•+; spectra from a typical set have been plotted after normalization at the first peak, in Figure 3 and labeled as curves 3-8. Unfortunately, these spectra cannot be subjected to a quantitative analysis for two reasons. In the first place, it was not possible to make a reliable correction for the absorbance of TFA at wavelengths shorter than 300 nm. Second, the original absorption spectrum of the sample (curve 1) changes, as soon as some TFA is added, to the spectrum labeled as curve 2, which is a superposition of a blue-shifted spectrum peaking at 288 nm and a small contribution, whose amplitude depends on the amount of TFA, arising from TOH•+; the 288 nm band may reflect the protonation of TOH, but, since this assignment has not been confirmed, we will designate the species responsible for the 288 nm peak as TOH?. We have also reproduced, in Figure 3, a spectrum (curve 9) assigned earlier to TOH•+ by Wilson and co-workers,15 who used electrochemical oxidation; it should be noted that their initial spectrum (before electrolysis) is also a superposition of the 288 nm peak, which they identified with TOH itself, and a small signal contributed by TOH•+. The visible band in our spectra (recorded in DCM/TFA at 295 K) is in excellent agreement with the corresponding band of curve 9 (recorded in DCM/TFA/Bu4NPF6 at 253 K); the slight discrepancy at the absorption onset, most likely due to the difference in temperature, will be ignored in this discussion. On the other hand, it is clear that the relative heights of our spectra at 300 nm are significantly smaller than that in curve 9; the difference must be attributed to the presence in the electrochemical cell, already noted by Williams and Webster,14 of species other than TOH•+ and TOH?. One compound that forms during the electrochemical oxidation experiments with a noticeable absorbance at 300 nm is the diamagnetic phenoxonium cation, TO+, which is longlived in dry MeCN.14,15 TO+ forms via the initially produced radical cation, TOH•+, deprotonating to form the neutral radical, TO•, which is then immediately oxidized at the electrode surface by another one-electron transfer. Even if one ignores the additional products and assumes that conversion from TOH to TOH•+ is quantitative, a critical look at the electrochemical spectra published in refs. 14-16 is needed before using them as a basis for quantitative deductions; we pause, therefore, to comment on these spectra. Occupied primarily with the electrochemical behavior of TOH and related compounds, the authors of ref 14-16 contented themselves with a visual examination of the spectral changes

J. Phys. Chem. A, Vol. 114, No. 40, 2010 10799

Figure 4. Curves 3-8 depict spectral changes following the addition of TOH to TFA; all spectra have been normalized to the visible peak of curve 3. Curve 2 is the spectrum of TOH?, normalized to have the same height as curve 1. For an explanation of curves 1 and 9, see Figure 3. The inset a provides a close-up view of the ultraviolet peaks of curves 3-8. In inset b, Ax is the absorption spectrum of TOH•+ obtained by subtracting the contribution made by curve 2 to curves 3-8.

during electrolysis. Unfortunately, they also overlooked the fact that the absorption coefficients of TOH in ref 15 and of two model compounds in ref 16 are too large by a factor of about three; when their spectra are rescaled, the absorption coefficients of the radical cations are found to be in the neighborhood of 5 × 103 M-1 cm-1. We have noticed, using singular value decomposition as well as the spectral reconstruction method, that a satisfactory simulation of their in situ spectra requires consideration of at least three species, two of which are TOH? and TOH•+; since the third spectrum is not known and the absorption coefficient of TOH? is unavailable, the electrochemical spectra cannot furnish the absorption coefficient of TOH•+. We therefore decided to follow the approach used by Depew and coauthors,4 who reacted TOH with TFA, and resorted to EPR spectroscopy for measuring the concentration of TOH•+ (relative to that of dpph•). They reported that thermal oxidation of TOH proceeded slowly, and the intensity of TOH•+ reached its maximum value only after about 12 h in an EPR tube sealed under vacuum; they also found that the reaction produced not only TOH•+ but also TO•, and the relative amounts of the two species depended on the initial concentration of TOH and on the sample treatment. We have worked with air equilibrated samples and used what they call “intermediate” concentrations (5 × 10-3 M); under these conditions, we observed only TOH•+, and its concentration reached a steady value more rapidly (within 1.5 h). A set of spectra was recorded during the growth and the steady state; one such set is plotted in Figure 4. The absorption coefficient of TOH•+ was found by using the relation sam Aref Iref p εp [dpph•] ) ) ref sam Isam [TOH•+] εp Ap

where I and Ap denote the doubly integrated area of the EPR spectrum and the absorbance at p, respectively, and the superscript specifies the absorbing entity. We were led to the x ) (1.58 ( 0.05) × 104 M-1 cm-1, which is in result ε461 reasonable agreement with the value, 1.82 × 104 M-1 cm-1, reported (without any error margin) in ref 4. Our attempts to generate TOAc•+ in the absence of other absorbing products did not yield satisfactory results; using the optical spectrum shown in Figure 5 as a guide, we have

10800

J. Phys. Chem. A, Vol. 114, No. 40, 2010

Naqvi et al.

Figure 6. In-situ absorption spectra obtained during the reduction of the mixture whose absorption spectrum is labeled as A∞ in Figure 5.

Figure 5. (a) In-situ absorption spectra obtained at 233 K in DCM containing 0.5 M Bu4 NPF6 during the sequential one-electron oxidation of 5.0 mM TOAc. (b) Plots of curves Ai (i ) 6, 7, ... 11) after normalization; the curve labeled A k0 is an amplified version of A0, plotted for the sake of easy comparison with the normalized spectra. xAc estimated that ε464 ) (1.29 ( 1.0) × 104 M-1 cm-1, which concurs with the value published in ref 13. Electrochemical Oxidation of TOAc. Cyclic voltammetry experiments indicated that TOAc could be electrochemically oxidized by one electron at approximately 1.0 V vs Fc/Fc+ in MeCN and DCM. The oxidation potential of TOAc is approximately +0.5 V more positive than that of TOH obtained under identical experimental conditions.14 In MeCN or DCM in the absence of added acid or base, the electrochemical oxidation of TOH occurs by the loss of two electrons (and loss of one proton) because the initially formed cation radical rapidly deprotonates and undergoes further electrochemical oxidation. Because TOAc does not contain an acidic phenolic proton, the cation radical formed by one-electron oxidation (TOAc•+), does not require the presence of acid (to stop deprotonation), unlike TOH•+ when it is generated by electrochemical means.14-16 Instead, TOAc•+ was produced by applying a potential >1.0 V vs Fc/Fc+ in pure DCM or MeCN (containing 0.5 M Bu4NPF6) in an in situ electrochemical cell. Because the in situ spectra have been recorded by using a scanning spectrophotmeter, the shape of a spectrum will be meaningful only if the spectral changes due to the reaction in progress are very slow compared to the time (2-3 min) needed for completing a scan. This condition is not fulfilled in the initial stages of the reaction, even at the low temperature employed here; after an initial rapid phase, the rate subsides, and it becomes justifiable to assume, and possible to verify the assumption, that each spectrum represents a practically timeinvariant composition of the mixture. A set of in situ spectra recorded during electrochemical oxidation of TOAc is shown in Figure 5a; the symbols A0 and A∞ will be used for denoting the initial spectrum (before electrolysis) and the final spectrum (recorded after 1.5 h, when no further change in the spectrum could be detected). The initial spectrum A0 is clearly contributed by TOAc. None of the intermediate spectra (Ai, i ) 1, 2, ... 11) could be reconstructed as a linear combination of A0 and A∞; to illustrate this remark, we have normalized Ai(i ) 7, 8, ... 11) to A∞ in the visible region and have plotted the normalized spectra along with A∞ and Aˆ0 (a magnified version of A0) in Figure 5b. The near-

coincidence of the normalized spectra in the visible region supports the view that the visible band arises from one species, which we take to be TOAc•+. Because A∞ provides no clue as to the fraction of the unreacted TOAc, and the contribution made by unidentified intermediates is unknown, we have no means of extracting the shape of the UV peak of TOAc•+ or estimating the absorption coefficient of TOAc•+. After recording the final spectrum, an attempt was made to recover the original compound by applying a reductive potential (Eappl ) 0 V vs Fc/Fc+); a second series of spectra was recorded during this stage, whose members will be denoted by A-r, with r ) 1, 2, ... etc. Figure 6 shows two such spectra along with two spectra from the first series (plotted in Figure 5). One can see that the relative height of the ultraviolet peak is larger in the second series, showing the accumulation of absorbing species different from TOAc and TOAc•+. Discussion Our main task in this section is to fathom the large discrepancy between the values of ε•p found in this work and the bulk of the results reported earlier.9-11 We note, among the preliminaries to our exposition, that a dual channel stoppedflow experiment consists in preparing two solutions at known concentrations and mixing them in a known volume ratio, say p:q. Thanks to the painstaking investigations of Mukai et al.,10,11 we now have a large body of relevant data presented in meticulous detail. Let us (i) use the symbol ∆ to denote the time needed for complete mixing, and ∆1 for the time at which the first observation is made; (ii) reduce eqs 4a,b to their model by putting k-1 ) 0 ) k2; and (iii) assume, for the sake of the argument, that the model is not at fault. Numerical integration of the resulting rate equations would predict the correct outcome of an experiment, provided that one has formulated an appropriate set of initial conditions, which requires fixing an origin of time (t ) 0). Now, this origin cannot be taken as the instant immediately before the start of the process in which solutions of TOH and frad• are mixed, even if one can guarantee, by controlling the experimental conditions, that E(0) and F(0) have the desired values and N(0) ) 0. This is because the rate equations come into force only after the dead time; in the intervening period, the concentrations of the reactants will change not only because of the reaction but also due to the mixing process itself; indeed, the concentrations will also depend on the spatial coordinates, since the composition of the mixture will become spatially homogeneous only at the end of the dead time. One can decide, of course, to use the rate equations for t > ∆, but then one must formulate new initial conditions at t ) ∆, since FΥ < F0 and NΥ > N0, where XΥ ≡ X(Υ) and 0
(1/2)A573 that ∆ > ∆1 in their apparatus, since t ) ∆ is defined as the earliest instant at which complete mixing may be taken for granted. We note next that S14(573) is close to zero (showing near-complete consumption of ArO•), and S14(423) ≈ 0.374. The contribution made by ArO• or ArOH to S14(423) is negligible, • N(t14)l. All we need which allows us to write S14(423) ) ε423 now is the value of N(t14), which can be equated, if we neglect the loss suffered by N during the short interval between t ) 0 and t ) t14, to (1/2)F0, with the factor 1/2 accounting for the 2-fold dilution due to mixing. The authors of refs 10 and 11 overlooked the dilution, and used instead N(t14) ) F0; in other • by a factor of 2. The reliability words, they underestimated ε423 • ) 7.64 × 103 M-1 cm-1, of the corrected value in toluene, ε423 will be assessed shortly. We have made use of Figure 2a in ref 11, for it is replete with spectral information, but the same conclusion is reached if one examines the kinetic trace in their Figure 2b, which shows that A423•, the absorbance of the mixture at 423 nm, reaches its maximum value (≈0.366) at t ) 53 ms. The conclusion that the absorption coefficients of TO• determined in refs 10 and 11 should be multiplied by a factor of 2 is now inescapable. When this is done for benzene and • • ) 6.90 × 103 M-1 cm-1 and ε427 chloroform, one obtains ε424 ) 8.64 × 103 M-1 cm-1, respectively; the corresponding data • ) published by Boguth and Niemann are (respectively): ε423 • 3 -1 -1 3 -1 -1 6.70 × 10 M cm and ε426 ) 8.11 × 10 M cm . For • ) DCM solutions10 the corrected value comes out to be ε427 8.20 × 103 M-1 cm-1, which compares favorably with our result (which is, after all, a lower limit). We conclude, therefore, that the corrected value for toluene solutions, derived above, is also • ) 5.0 × reliable. For hexane,10 even the corrected value, ε418 103 M-1 cm-1, is appreciably smaller than our result and with the values found in compliant solvents; this discrepancy requires further investigation. Solving the rate equations is necessary only when one wants to verify that the chosen model fits the observed kinetic traces. If one is content with the choices k-1 ) 0 ) k2 and E0 . F0, it is much simpler to (i) discard the rising part of A•p, the absorbance signal representing N ≡ TO•, (ii) extrapolate the truncated signal to t ) 0 (the instant prior to mixing), and (iii) replace the original rate equations by those stated below (we still have k-1 ) 0 ) k2):

F)0

(6a)

dN ) -k3N2 dt

(6b)

which are to be solved subject to the initial condition

N0 )

p F p+q 0

(7)

The above prescription, valid when frad• ) ArO•, needs a straightforward amendment when frad• ) dpph•, since absorption by dpphH is not negligible at wavelengths where TO• has its first aborption peak. Those who have used eq 6b and dpph• include Boguth and Nieman8 and the authors of • in ethanol happens to be 3.8 × ref 9, whose value for ε426 103 M-1 cm-1, about half the expected value, which is probably fortuitous, since they used neither a stopped-flow apparatus for this part of their work nor a dilution factor of 2. Since we have not been able to infer the initial conditions used in these publications, we are inclined to the view that Boguth and Niemann must have used the correct condition (without explicitly stating it), but the authors of ref 9 did not, and if they did, their low value must be attributed to some other error(s). Having reconciled the discrepancy between the pioneering studies of Boguth and Niemann and the more recent calculations based on a carefully conducted experimental study of Mukai and coauthors,10,11 we conclude that ε•p ) (7.5 ( 0.5) × 103 M-1 cm-1 represents a reliable typical value. Our method for finding the value of ε•p, devised to bypass kinetic modeling, is based on a robust global analysis of timeresolved spectra recorded after mixing the reactants manually. Removal of moisture and dissolved oxygen is crucial for obtaining consistent results; values of the ratio ε•p/ε•d p significantly smaller than 0.6 were occasionally obtained, but such results were regarded as spurious and were attributed to the presence of adventitious reactants. It can hardly be overemphasized that, since 0.6 represents, according to our experiments, the lower limit for the ratio in hexane and DCM; a better experiment (with a shorter dead time or performed with purer solvents) cannot yield a smaller value. Acknowledgment. The authors thank the Research Council of Norway for financial support (Project No. 191102). References and Notes (1) Burton, G. W.; Ingold, K. U. Acc. Chem. Res. 1986, 19, 194. (2) Traber, M. G.; Atkinson, J. Free Rad. Bio. Med. 2007, 43, 4. (3) Doba, T.; Burton, G. W.; Ingold, K. U.; Matsuo, M. J. Chem. Soc. Chem. Comm. 1984, 461. (4) Depew, M. C.; Craw, M. T.; MacCormick, K.; Wan, J. K. S. J. Photochem. Photobiol., B 1987, 1, 229. (5) Tasinato, A.; Boscoboinik, D.; Bartoli, G. M.; Maroni, P.; Azzi, A. Proc. Natl. Acad. Sci. U.S.A. 1995, 92, 12190. (6) Naqvi, K. R.; Melø, T. B.; Sliwka, H.-R.; Mohamad, S. B. B.; Partali, V. Photochem. Photobiol. Sci. 2003, 2, 381. (7) Naqvi, K. R.; Melø, T. B.; Ja´vorfi, T.; Gonza´lez-Pe´rez, S.; Arellano, J. B. Phy. Chem. Chem. Phys. 2009, 11, 6401. (8) Boguth, W.; Niemann, H. Int. Z. Vit. Forsch. 1969, 39, 429. (9) Gregor, W.; Grabner, G.; Adelwo¨hrer, C.; Rosenau, T.; Gille, L. J. Org. Chem. 2005, 70, 3472. (10) Nishioku, Y.; Ohara, K.; Mukai, K.; Nagaoka, S. Bull. Chem. Soc. Jpn. 2009, 82, 494. (11) Ouchi, A.; Ishikura, M.; Konishi, K.; Nagaoka, S.; Mukai, K. Lipids 2009, 44, 935. (12) Svanholm, U.; Bechgaar, K.; Parker, V. D. J. Am. Chem. Soc. 1974, 96, 2409. (13) Sur, S.; Colpa, J. P. Chem. Phys. Lett. 1986, 127, 577. (14) Williams, L. L.; Webster, R. D. J. Am. Chem. Soc. 2004, 126, 12441. (15) Wilson, G. J.; Lin, C. Y.; Webster, R. D. J. Phys. Chem. B 2006, 110, 11540. (16) Peng, H. M.; Choules, B. F.; Yao, W. W.; Zhang, Z. Y.; Webster, R. D.; Gill, P. M. W. J. Phys. Chem. B 2008, 112, 10367. (17) Mori, T.; Izumi, H.; Inoue, Y. J. Phys. Chem. A 2004, 108, 9540.

10802

J. Phys. Chem. A, Vol. 114, No. 40, 2010

(18) Nakanishi, I.; Fukuhara, K.; Shimada, T.; Ohkubo, K.; Iizuka, Y.; Inami, K.; Mochizuki, M.; Urano, S.; Itoh, S.; Miyata, N.; Fukuzumi, S. J. Chem. Soc. Perkin Trans. 2 2002, 9, 1520. (19) Naqvi, K. R.; Melø, T. B.; Raju, B. B. Spectrochim. Acta A 1997, 53, 2229. (20) Naqvi, K. R.; Hassan, T. H.; Naqvi, Y. A. Spectrochim. Acta A 2004, 60, 2783.

Naqvi et al. (21) Friaa, O.; Chaleix, V.; Lecouvey, M.; Brault, D. Free Radical Biol. Med. 2008, 45, 1011. (22) Friaa, O.; Brault, D. Org. Biomol. Chem. 2006, 4, 24. (23) Musialik, H.; Litwinienko, G. Org. Lett. 2005, 7, 4951. (24) Foti, M. C.; Daquino, C.; Geraci, C. J. Org. Chem. 2005, 69, 2309.

JP106736X