which solvent quenching can be bypassed by exciting only the fluor. Such data have not been obtained for bisMSB and dimethyl-POPOP. Previous spectrofluorometric studies, however, showed that those compounds which caused substantial quenching of other long wavelength fluors did so only at concentrations which were prohibitively high for liquid scintillation counting, because of their concurrent severe solvent quenching. Although some compound may be encountered which will severely quench one of these secondaries and alter its efficiency relative to the others, this is expected to occur only rarely. Because in many counters the principal usefulness of a secondary solute is in compensating for direct quenching of the primary solute, the three secondaries were compared for effectiveness in samples containing one of three compounds known to quench CCl,, 4-hydroxy-4-methyl-2PPO : pentanone, or PPO itself at 30 grams/ liter. This latter is particularly important because the optimal primary
solute concentration may be quite high in the presence of other quenching agents. Each secondary was used a t or near its maximal concentration a t 4’ C.: POPOP a t 0.6 gram/liter, bis-MSB at 2 grams/liter, and dimethyl-POPOP a t saturation (a precipitate formed when samples a t 2 grams/liter were cooled to 4’ C.). With each quencher, efficiency was improved by the secondaries in the order bis-MSB > dimethyl-POPOP > POPOP. Tritium efficiencies in samples with bis-MSB were 1.1to 1.2 times those in POPOP samples, while dimethylPOPOP samples were 1.03 to 1.07 times more efficient than POPOP. The relative efficiencies vary somewhat with the kind and degree of quenching, but the values observed in these experiments are indicative of the order of magnitude to be expected. When comparisons were made a t equal concentrations by weight of secondary solute, efficiencies with POPOP were 1.01-1.03 times those with bis-MSB in samples with 30 grams/liter PPO, and were 1.08-1.18 times those with dimethyl-POPOP with each of the
quenching agents. The order POPOP > bis-MSB > dimethyl-POPOP is the same as for unquenched samples. The chief advantage of the new fluor, bis-MSB, is its higher solubility. It should find its greatest usefulness a t high concentrations in samples in which the primary fluor is quenched by some other sample component or through selfquenching. LITERATURE CITED
(1) Bush, E. T., Hansen, D. L., “Radio-
isotope Sample Measurement ‘Techniques in Medicine and Biolo p. 395, International Atomic %kergy Agency, Vienna, 1965. (2) Birks, J. B., “The Theory and Practice of Scintillation Counting,” p. 31615, Macmillan, New York, 1964. (3) Hayes, F. N., et al., U. S. At. Energy Comm. Rept. LA-1639. (19531. (4) Ott,. D. G., “Liquid Scintillation Counting,” C. G. Bell, F. N. Hayes, eds., p. 101-7, Pergamon Press, Oxford, 1958. ELIZABETH T. BUSH Nuclear-Chicago Corp. 333 E. Howard Ave. Des Plaines, Ill.
The Rate of Oxidation of Platinum Electrodes SIR: Feldberg, Enke, and Bricker (2) have reported that the oxidation
charge accumulated a t constant potential in the anodic oxidation of platinum in perchloric acid is proportional to the logarithm of oxidation time over a wide range of time and applied potential. Quite similar results were obtained by Smith (6) in sulfuric acid solution. The first authors have given an interpretation of this behavior by writing a conventional electrochemical rate equation and making the assumption that the forward rate constant decreases with increased oxidation of the surface in an exponential manner. The origin of this Temkin-like behavior is uncertain, but the introduction of an exponential term of this kind appears to be justified by experimental data. Such a treatment implies that the slope of the oxidation charge us. log time curve should be independent of potential at high oxidation potentials. At lower oxidation potentials, the curves are less linear, and the slopes depend upon potential. The work reported here attempts to extend the treatment to lower potentials. It is well known that the nature of an anodically oxidized platinum electrode is not completely specified by a specification of the oxidation charge present upon the electrode, but rather, depends strongly upon the time and potential history leading to the formation of the oxide charge. The hysteresis in the 1242
ANALYTICAL CHEMISTRY
anodic and cathodic curves for platinum oxidation and reduction may be cited (3). The magnitude of charge capable of existing on smooth platinum electrodes corresponds in many cases to nine or ten electrons per surface platinum atom, if a roughness factor not much greater than one is taken. This large charge cannot be accommodated by any reasonable oxidation state of surface atoms alone. The fact that oxygen atoms readily reach positions below the surface has been demonstrated by low energy electron diffraction (7) and by field ion microscopy (6). Recently, Warner and Schuldiner have used the term “dermasorption” for such subsurface oxides (9). It is proposed here that the exponential decay of the rate constant may be taken as the limiting form in a two-step process, the first of which is regarded as an electrochemical reaction a t the surface. This is followed by a potential-independent transfer of oxide to the underlying platinum structure. Alternatively, film growth by metal atom transfer may be involved. The two steps are quite different from those discussed by Feldberg, Enke, and Bricker ( 2 ) . EXPERIMENTAL
Apparatus. The circuitry used for constant potential electrolysis and integration of charges accumulated was similar to that previously descri.bed (1).
Electrodes. Two types of test electrodes were used. A single crystal platinum electrode approximately 2 mm. in diameter was mounted in the end of a piece of glass tubing with black sealing wax. This electrode had an exposed geometric area of 0.091 cm.2 The second electrode was polycrystalline platinum wire sealed into soft glass tubing. After annealing, this electrode was polished using kerosene-levigated alumina supported on a felt disk mounted on a dental engine. The exposed geometric area was 0.110 cm.2 Electrochemical measurements made with the single crystal were found to be the same as those made with the wire electrode within experimental uncertainty, after allowing for differences in area. Most of the data reported here were obtained with the wire electrode. Cell and Electrolvte. All electrolyses were made h 0.5M H$Oc prepared by diluting reagent grade acid with water distilled from alkaline permanganate in a borosilicate glass still. The electrolyte was freed of oxygen by purging with nitrogen obtained from evaporation of liquid nitrogen and passed over heated copper. The working electrode of platinum gauze in 0.5M H2S0, was in a compartment separated from the test electrolyte by a glass frit. Potentials were measured against a saturated calomel electrode connected by a capillary filled with 0.5M H2S04. Potentials are herein referred to saturated calomel. Stirring the electrolyte had no effect upon the charges observed;
action. The second step, transfer of charge to underlying structure, may be written 1.6
where $ is the charge stored in subsurface layers. Here the reverse process can be ignored. The form of Equation 3 is justified only by the observed behavior of platinum a t high potentials and long times. The change of surface coverage, then, is
1.2
.
N
i 0 Y)
m
z
de - = dt
s
g 0.8 -I W
e I
0.4
0.0
Oxidation charge vs. time
Circles, experimental; lines, computed
electrolvses were made in unstirred solution. Electrode Pretreatment and Storage. Electrodes were stored in 0.5M H2S04 under continuous nitrogen purging and kept a t + 0 . 2 volt us. S.C.E. It was necessary to treat the electrode in this way for several hours after anodization, t o obtain reproducible results. No other treatment was used. RESULTS AND DISCUSSION
Cyclic Potential Experiments. Starting from a well-reduced electrode, a one-second anodic pulse of constant potential was applied. The potential was then switched to +0.20 volt for one second, then a second one-second anodic pulse was applied, and the sequence continued many times. The charge accumulated was continuously recorded. In separate experiments, anodic potentials of from +0.70 to +0.90 volt were used. The charge consumed a t anodic potential was always greater than that released a t +0.20 volt. Both charges were transferred very quickly, the current dropping to a very low value in less than the one-quarter second response time of the recorder. Thus there was a fairly rapid oxidation charge, most of which could quickly be recovered. Such charges increased with increasing number of cycles, reaching a steady value in from ten t o two hundred cycles, and were greater a t higher potentials. Such charges were several
hundred microcoulomb cm.-2, too large to be accounted for solely from double layer charging. I n addition to the recoverable charge, the total charge increased with cycling a t an ever diminishing rate but did not approach any steady value a t several hundred cycles. A quantitative treatment of the above cannot be given, but the results suggest that the oxidation of platinum be considered as a two-step process, the first of which is rapid and easily reversible, and is followed by a slow, irreversible step. Several hours storage a t +0.20 volt was required for complete removal of the charge stored in the slow process. Anodic Oxidation at Constant Potential. Of the many two-step formulations which could be written, we have chosen one which we believe to be as simple as possible and yet reflect the above observations. Assuming the activity of the oxidized part of the surface is proportional to the fraction of the surface e,&,, we write for the easily reversible surface reaction
rf
=
k(e,
- 0 ) exp anF(E
?b
=
- Eo)/RT
(1)
k0 exp
l)nF(E - E")/RT (2) where e is surface charge and eo is surface charge a t complete coverage, k is the heterogeneous rate constant, and Eo becomes the potential a t which a half-covered surface is in equilibrium with respect to the electrochemical re(a -
- rb - G!?
dt
(4)
Total oxidation charge (e+$) can be obtained by simultaneous integration of Equations 3 and 4, taking the initial charge as zero. These equations were integrated numerically for a variety of parameters, using an International Business Machines Type 650 computer. The equations include a large number of parameters, but certain of these can be evaluated independently, or reasonable values assumed. A t long times, e should approach an equilibrium value given by equating Equations 1 and 2. Under these circumstances, Equation 3 may be integrated, with the result
2
I
Figure 1 .
T/
$c = ln(1
+ acet)
(5) Thus, c may be calculated from the slopes of the charge us. In time curves a t long times. For the electrode used here, we find c = 0.0038 microcoulomb-1 cm.2, the same as found by Smith (6) for a different electrode, but much smaller than that calculated from the data of Feldberg, Enke, and Bricker [(Z), Figure 121. We assume Bo = 960 microcoulomb cm.-* for true surface area, four electrons per surface platinum atom; and a roughness factor of 1.15, essentially that of Laitinen and Enke (3). It can be seen that the ratio, ke,/a, strongly determines the shapes of the calculated oxidation charge us. loglo time curves, but the absolute values simply determine the positions of the curves on the time axis. A large number of curves were computed using various values of these parameters, with the result that curves of the approximate shapes of the experimental curves could be obtained only if this ratio were taken close to unity. The curves in Figure 1 were obtained using k = 1 see.-' and a = 1000 microcoulomb sec.-l cmaF2 These values imply an exchange current roughly equal to the minimum exchange current needed for polarographic reversibility for dissolved reactants a t about millimolar concentration. A value of CY = 0.5 has been used, since greatly different values give curves spaced too closely with respect to potential either above or below E'. The parameters, VOL. 38, NO. 9, AUGUST 1966
1243
E o and
n, can be adjusted to fit the computed curves to the experimental data with respect to potential, once curves of the proper shape and scale have been obtained. For the curves in Figure 1 we have taken Eo = +l.OO volt us. S.C.E., but found that n must be taken as approximately 0.5 to obtain a proper spread in charge with respect to potential. We would have preferred to take n = 1 to reflect a stepless removal of protons from the solvent sheath on the electrode. Vermilea (8) has found that, in the oxidation of tantalum, the activation energy of the primary process is not a simple linear function of potential and that the preexponential factor is a function of potential. Use of these ideas in formulating different forms for Equations 1 and 2 would permit a better fit with a more reasonable value of n. Similar effects could be obtained by allowing a to vary with coverage. Such modifications would introduce additional arbitrariness not justified by available data. No allowance for double layer charging has been made in the computed curves, but this is a small error. It is not intended to suggest that the above parameters are the correct ones.
In particular, it appears that the formulation of surface activity is quite in error. It is suggested, however, that a two-step model involving a nearly reversible surface reaction, followed by an exponential transfer to underlying layers is capable of explaining much of the kinetics of electrochemical oxidation of platinum. It provides for the storage of large amounts of charge in the subsurface form and partially shielded from the influence of applied potential, hence the hysteresis in the reduction of such surfaces; and for the observation that the total charge does not determine the nature of the surface. In this connection, the above computations show large fractions of the total charge in the subsurface form, even a t low potentials and short times. It is of interest that equations of the form of Equation 3 have been used to treat the growth of very thin films by place exchange of metal and oxygen atoms (4). It should be noted that results similar to, but not identical with, those obtained experimentally above were obtained if charges were measured by cathodic stripping of repeatedly oxidized electrodes, rather than by direct oxidation of well-reduced electrodes,
LITERATURE CITED
(1) Booman, G. L., Morgan, E., Crittenden, A. L., J . Am. Chem. SOC.78, 5533
(1956). (2) Feldberg, S. W., Enke, C. G., Bricker, C. E., J. Electrochem. SOC.110, 826 (1963): (3) Laitinen, H. A., Enke, C. G., Ibid., 107, 773 (1960). (4) Lanyon, M. A. H., Trapnell, B. M. W., Proc. Roy. SOC.A227, 387 (1955). (5) Muller, E. W., “Structure and Properties of Thin Films,” C. A. Neugebauer, J. B. Newkirk, D. A. Vermilea, eds., p. 476, Wiley, New York, 1959. (6) Smith, J. G., Thesis, University of Washington, Seattle, 1960. (7) Tucker, C. W., J . A p p l . Phys. 35,1897 (1964). (8) Vermilea, D. A., J . Electrochem. SOC. 102, 655 (1955). Zbid., (9) Warner, T. B., Schuldiner, S.,’ 112, 853 (1965).
KARLH. POOL’ JAMES G. SMITH$ A. L. CRITTENDEN Department of Chemistry Universit of Washington Seattle, $ash. 98105 1 Present address, Department of Chemistry, WFhington State University, Pullman, Wash. 2 Present address, Pennsalt Chemicals Corporation, King of Prussia, Pa.
Quantitative Determination of Individual Tocopherols by Thin Layer Chromatog ra p hic Se pa ration a nd S pectro photo metry SIR: The natural antioxidants such as the tocopherols are important factors affecting the shelf life of vegetable oils. The fact that the various tocopherols differ considerably both in their biological activities and in their ability to protect these oils from oxidative degradation emphasizes the need for good analytical methods. The high molecular weight and chemical similarity of the various tocopherols have hindered their efficient separation. I n the past few years a number of workers have reported thin layer chromatographic systems for the separation of the tocopherols (1, 2, 7-18). Several analytical methods have been reported which utilize such separation. Seher (10) determined the tocopherols semiquantitatively from the area of the spot obtained on the thin layer plate. Dilley and Crane (8) determined the tocopherol content, after thin layer separation, from the absorbance difference obtained after a two-step procedure involving oxidation followed by reduction. Katsui, Ichimura, and Nishimoto (7) analyzed only standard a-tocopherol solutions using the Emmerie-Engel reagent. The method described here was developed for the analysis of tocopherols in peanut oil and involves saponification 1244 *
ANALYTICAL CHEMISTRY
of the oil sample and the separation of the tocopherols in the nonsaponifiable fraction by thin layer chromatography. The CY-, y-, and &tocopherols were removed from the thin layer plate and determined spectrophotometrically by the Tsen (13) modification of the Emmerie-Engel method for total tocopherols (3) using bathophenanthroline. This modification increases the sensitivity of the original EmmerieEngel procedure 2.5 fold. The method was applied to peanut oil samples and the precision of the method was evaluated. EXPERIMENTAL
All reagents were of analytical grade. Bathophenanthroline was purchased from the G. Frederick Smith Chemical Co.; a-tocopherol was obtained from Merck and Co., Inc.; p-, y-,and &tocopherol were obtained from Hoffmann-LaRoche, Inc. The bathophenanthroline reagent was 6.0 X lO-3M in ethanol and was stored in an amber bottle in the refrigerator. The ferric chloride solution was 1.0 x 10-3M in ethanol, and was prepared fresh each day and protected from light by an amber bottle. Orthophosphoric acid solution was 0.1M in ethanol; pyrogallol was 5% in ethanol. Potassium hydroxide solution was preReagents.
pared by dissolving 160 grams in 100 ml. of distilled water. All distilled water used was boiled to remove dissolved oxygen. Potassium ferricyanide and ferric chloride were 5% in distilled water. Mallinckrodt anhydrous ether was found preferable; others gave irregular results due to peroxides. Saponification. One gram of peanut oil, 4 ml. of 5% pyrogallol in absolute ethanol, and 20 ml. of additional ethanol in a three-neck, lOO-ml., round-bottom flask were brought to reflux under a NP atmosphere. One milliliter of KOH (160 grams/100 ml. distilled HzO)was quickly added, and refluxing continued for 1 hour. The reaction mixture was cooled by immersing the flask into an ice bath without removing it from the reflux condenser or Nz atmosphere. Boiled, cooled, distilled water (20 mi.) was then added while the mixture was still under Nz. The flask was then removed from the condenser and Nz atmosphere and quickly stoppered. After the reaction mixture had dissolved in water, it was extracted five times with fresh anhydrous ethyl ether. The ether extract was kept under NZ until the extraction was completed. The extract was then washed with boiled, cooled, distilled water, until free of base. The ether was evaporated, and the dry nonsaponifiables were dissolved in 5 ml. (VI) of ethanol. Four milliliters ( V z ) was removed to a 20-ml. round-
