H. C. A.
1704
characteristic of the adsorbate. It is then stated that, “the scattering indices define the oscillator strengths per unit area, it is reasonable to assume that they are proportional to coverage,” so that ui = 6u’i and # = $ ezptiUti. To put this into the notation previously used, suppose measurements to be made at the polarizing angle. Then IJ = 0 and # = p , so that p = 6ZP’iu’i. At 6 = 1, p = ZPiui so that p = @, the linear dependence assumed by 0thers.l’ However, if each adsorbed molecule constitutes an oscillator, then the scattered intensity depends upon the number of molecules, i e . , on the surface coverage. The intensity, in turn, appears to depend upon ui2, for ui defines an amplitude, not an intensity. Then O a ui2, so that p =
+
VAN
BEERAND P. M. HEERTJES
pO’la, which is identical with the equation derived
via
the Fresnel-Drude laws.
Conclusion The Fresnel and Drude equations, known to be valid in their spheres of application, imply that for surface coverages less than a monolayer (e < 1) the dependence of ellipticity upon surface coverage at the polarizing angle is p = p e l / ’ . Previous investigators have assumed that p = PO, which seems not to be true. The limits of the derived expression remain to be established by accurate experiments.
Acknowledgment. The author thanks Socony Mobile Oil Co., Inc., for permission to publish this work.
The Mechanism of the Photoreduction of Azo Dyes in the Presence of DL-Mandelic Acid and in the Absence of Oxygen
by H. C. A. van Beek and P. M. Heertjes Laboratory of Chemical Technology, Technical University, Delft, Holland
(Received July 14, 1066)
The photoreduction of azo dyes in solutions containing DL-mandelic acid has been studied under anaerobic conditions using light of 253.7 mp. From the results it appears that the reaction is caused by a strong reducing species formed upon absorption of light by DLmandelic acid. The photoexcited DL-mandelic acid is deactivated in at least two different reactions, one of which results in the production of an active reducing species. In absence of azo dyes (or other oxidizing agents) the formation of the active reducing species is reversible. It is possible that this reducing agent is identical with the triplet state of DL-mandelic acid. The quantum efficiency of the formation of the active reducing species has been determined and is a suitable measure for the activity of the substrate (DL-mandelic acid) as a hydrogen donor.
Introduction In a previous paper,’ the fading of azo dyes with po]ychromatic light in the presence of different substrates has been described. It has been found that upon photoexcitation many substrates Can act powerful reducing agents. Irradiation of solutions of The Joumcal of Physical Chemietry
such substrates and azo dyes in water under anaerobic conditions results in rapid photoreduction of the dyes to the corresponding amines. The wavelengths of the (1) H.C. A. van Beek and P. M. Heertjes, J . soc. Dyers Colourists,
79, 661 (1963).
PHOTOREDUCTION OF Azo DYES
light which is photochemically active in these systems are restricted to the region of wavelengths absorbed by the hydrogen donor (substrate). A mechanism for the fading of azo dyes in solution in the presence of active hydrogen donors and in the absence of oxygen has been proposed.' (Oxygen inhibits the reaction strongly.) It has not been possible, however, to draw conclusions regarding the nature of the active species which is formed upon photoexcitation of the hydrogen donor. Also, the effect of the concentrations of the reacting agents on the quantum efficiency of the reaction have not been determined, since this effect could not be separated with sufficient accuracy from the effect of the relative absorption of the reacting agents at the different wavelengths of the polychromatic light used. New experiments have therefore been carried out in which the anaerobic fading of different azo dyes in solution, initiated with monochromatic light, has been investigated. DL-Mandelic acid as a hydrogen donor was chosen because of its good absorbance at the wavelength (253.7 mp) of the main emission line of the lamps used (Philips TUV, 6 w) and its low absorbance at lines of other wavelengths also emitted by the light source.
Experimental Part The relative spectral energy distribution of the lamps used and the absorption spectrum of DL-mandelic acid are given in Figure 1. The radiation of 313.0 mp and higher wavelengths is only slightly absorbed by Dbmandelic acid. The light absorbed by the specific systems under investigation can therefore be considered as monochromatic. With the aid of the extinction coefficients of the substances concerned, inner filter factors can be calculated and used in the determination of the quantum yields of the photoreduction of the azo dyes. The fading experiments were carried out in the apparatus shown in Figure 2. Solutions of the dyes and DL-mandelic acid in water were irradiated in the quartz cuvette V with light from four (only two shown) low-pressure mercury vapor lamps (Philips TUV, 6 w) placed around the cuvette. When operated at constant voltage (220 v), the intensity of the radiation emitted by these lamps is constant over a long period of time. The amount of energy absorbed by the solutions was therefore linear with the time of exposure. The rate of light absorption by the solutions in cuvette V (diameter 33 mm, height 60 mm, optical density of all solutions at 253.7 mp >2) was determined with a ferric oxalate actinometer solution.2 In all calculations the quantum efficiency of the formation
1705
lcgE of mandelic acid
I I
I I
250
300
350
400
450
500
.
550
Figure 1.
Figure 2.
of Fez+ ions at 253.7 mp was taken as 1.25. From the lamp data (provided by the manufacturer, Philips, Eindhoven, Netherlands) and the result of the actinometric measurements it was calculated that the rate of light absorption of wavelength 253.7 mp in the cuvette was 1.8-1.9 X 10-6 einstein/min. Oxygen was removed from the solutions by passing through nitrogen gas for 30 min before starting the irradiation, The nitrogen gas (purity 99.997%, oxygen content
+ +C-COOH \
OH
+CH=O
OH
-cot -C--
(-1
J%
+C-COOH I/
I/ 0 where the asterisks denote a photoexcited molecule. Apparently, DL-mandelic acid is oxidized a t the a-carbon atom. A similar result has been obtained by Wells,6 who found that alcohols were oxidized by photoexcited quinones on carbon atoms a t the a position to the hydroxyl group. It must be noted, however, that in the latter case the photoexcited quinone is the active oxidizing agent, whereas in the case of photoreduction of azo dyes or oxygen the photoexcited DL-mandelic acid forms the active reducing species. I n the scheme presented, it is assumed that phenylglyoxalic acid is an intermediate in the photooxidation of DL-mandelic acid. No attempt has as yet been made to detect this in irradiated solutions. Also, the radical (I) formed by hydrogen abstraction of DLmandelic acid could enter into a dimerization reaction resulting in the formation of a polybasic acid. Such a reaction, however, must be of minor importance in this system, since it has been shown that benzaldehyde is the main reaction product. I n the scheme, the fact that irradiation of solutions of DL-mandelic acid in water in the absence of oxidizing agents with light of 253.7 mp causes no irreversible change has been accounted for. KO information is a t present available on the nature and properties of the excited states of DL-mandelic acid. The reactions described in this paper, however, indicate that photoexcit,ation of DL-mandelic acid re-
Figure 5.
-
5
0
7
%
&.lo'
sults in the reversible formation af an active reducing species (HX'). If oxidizing agents are present in the solution an irreversible conversion takes place. The system can be represented by the reactions
HX h", HX
(0)
HX* -% HX
(1)
-% HX' HX' -% HX* HX' -% HX + Q --%QH. + x. HX*
HX'
(2) (3)
(4)
(5) where HX* is photoexcited DL-mandelic acid, and HX' is the active reducing species. It can be assumed that the contribution of reaction 3 to the formation of the photoexcited hydrogen donor molecules is small compared to the direct formation of HX* by light absorption and can be neglected. If the fraction of light absorbed by HX is given by the inner filter factor (F) the fading rate of the dye expressed in moles/einstein is given by
-dQ - - Fdn
ks [Q1
k2
ki
+
k2
(kg
+ k d + a[Ql
(11)
(6) c. F. Wells, Trans. Faraday SOC.,57, 1703 (1961).
Volume 70, Number 6 June 1966
1710
H. C. A.
or
At high values of [HXIomost of the light entering the solution is absorbed by DL-mandelic acid ( F = 1). From the results given in Figures 4 and 5 it follows that in this case for the seven different dyes used the same values of the fading rates are found (M-1). Since it can be expected that the rate constant (k5)for reaction 5 is different for the different dyes, this indicates that in all cases k5[&]0 >> k3 kq. Apparently, reaction 5 is much faster than reactions 3 and 4 in the system investigated. Because no appreciable change in the rates has been found in the first period of the fading reactions, it also may be concluded that for the concentration range of the dyes k[&]>> k3 k4. Therefore, the rates of fading are mainly determined by the inner filter effects. At high values of [HXIochanges in the absorbances of the dye and of the reaction products only have a very small influence on the inner filter factor F (see eq IV). In this region the average fading rates approach the initial fading rates. From M-1, therefore, the quantum efficiency of the hydrogen transfer from DL-mandelic acid to the azo dyes can be calculated and has an average value of 0.20 g-atom of H transfered/einstein absorbed (variation 0.19-0.24). At lower values of [HXIo or at higher values of [&lo/ [HXlo the influence of the inner filter factor becomes noticeable. Since reaction 5 is very fast, eq I11 can now be transformed into
+
+
_ _ _1- _ --- 1 ki dQ/dn
F
+ kz kz
From this equation, it can be seen that the fading rate will decrease with increasing conversion of the dye if E, > E,. The accuracy of the determination of the fading rates, however, was not high enough to measure this effect for the dyes concerned. All fading rates are average fading rates. From eq IV it can be derived that a t high values of [&]O/[HX]othe differences between the initial fading rates and the average fading rates are greater than a t low values of [&Io/ [HXIo because in the latter case the total optical density is mainly determined by E , [HX]. For the initial fading rates the eq IV can be written as
The Journal of Physical Chemistry
VAN
BEEKAND P. M. HEERTJES
Comparison of eq V with eq I shows that if initial fading rates were used in eq I, N should be equal to Eo/E.. As has been shown above, it can be expected that the measured average fading rates are appreciably smaller than the initial fading rates a t high values of [&Io/ [HX lo. It follows that the substitution of the average fading rates in eq I leads to values of N which are higher than Eo/E, in the cases that E, > E,. For the dyes investigated, the direction of the deviation is as expected. The conclusion can therefore be drawn that the influence of the ratio of the initial dye concentration and the concentration of DL-mandelic acid is caused by the fact that with increasing value of this ratio a greater part of the light entering the reaction vessel is rendered inactive by the absorption by the dye. Also, it may be concluded that for all the experiments described, the same quantum yield is found for the hydrogen transfer reaction if the fading rates are corrected with respect to the inner filter effect. Since the value of the quantum efficiency is
