Anodic processes on n-titania with and without illumination in

Oct 1, 1988 - Anodic processes on n-titania with and without illumination in trifluoromethanesulfonic acid monohydrate saturated with alkanes...
0 downloads 0 Views 1MB Size
5716

J . Phys. Chem. 1988, 92, 5716-5721

Anodic Processes on n-TiO, with and without Illumination in Trifiuoromethanesulfonic Acid Monohydrate Saturated with Alkanes Yutaka Harima* and S. Roy Morrison Department of Physics, Simon Fraser University, Burnaby, B.C., Canada, V5A IS6 (Received: August 25, 1987; In Final Form: March 11, 1988)

Current doubling was observed on illuminated n-Ti02 electrode in trifluoromethanesulfonic acid monohydrate containing saturated hydrocarbons such as propane. The observation shows a successful photoelectrochemical oxidation of these difficult-to-oxidize agents by holes together with a generation of a highly reducing intermediate species, presumably an alkane radical. With no alkane, a gaseous product of the photooxidation is observed, but when propane is added the product is nongaseous. One of the final products of the propane oxidation is a white and electrically insulating polymer originating from alkane and solvent molecules. In contrast to the alkane oxidation by the photogenerated holes, solvent oxidation is the sole anodic reaction on a heavily doped n-TiO, in the dark that is reverse-biased to “breakdown”. In addition, corrosion of the semiconductor material was found to be appreciable on the illuminated n-Ti02 but not on the reverse-biased n-Ti02 in the dark. The differences in behavior between the two electrodes were explained by assuming different mechanisms for primary charge-transfer steps on the n-Ti02 under illumination and in the dark.

Introduction In an earlier note,’ we reported that propane oxidation gives a current doubling at an illuminated n-TiO, in trifluoromethanesulfonic acid (triflic acid) monohydrate at temperatures around 100 OC. This observation means first that a saturated hydrocarbon of a low molecular weight can be oxidized by photogenerated holes from a valence band of n-TiO,, although on metal electrodes, in general, low molecular weight alkanes are difficult to oxidize. The difficulty in general is their highly anodic oxidation potentials, which are beyond the oxidation potentials of the solvents used. The observation also means that the partially oxidized intermediate species has a relatively long life as well as a strong reducing power, allowing an electron to be injected into the n-Ti02 conduction band. The observation of current doubling appears to be of great importance in view of the fact that it ensures a formation of a highly reactive intermediate that can be useful in electrosynthesis reactions. In this paper, we describe current doubling with some alkanes, especially propane, a t the illuminated n-TiO, in more detail and describe anodic processes on a reverse-biased n-Ti02 in the dark. This report constitutes a part of our program aiming at the control of the generation of reactive species a t semiconductor electrodes and the control of their subsequent reaction with other species. The objective is synthesis using simple compounds such as saturated hydrocarbons as starting materials. Experimental Section Reagents. Triflic acid monohydrate was prepared by mixing 100 g of triflic acid (3M Corp.) and 18 g of distilled watet followed by a triple distillation of the mixture under argon atmosphere. The acid monohydrate, a white solid at room temperature, (mp 34 oC),2 was stored in a Schlenk tube filled with argon. When used at higher temperatures, the solvent had a high conductivity and no supporting electrolyte was required. Saturated hydrocarbons from methane to n-butane were instrument grade (99.5% minimum purity) from Matheson and were used, except for nbutane, after being passed through a cold trap kept at ca. -30 O C to remove impurities. Spectrograde n-hexane was purchased from Caledon and used without further purification. Argon and oxygen from Linde were prepurified grade and zero gas grade, respectively. Flow meters used for mixed gas experiments were calibrated with the respective gas prior to use. Sample Preparation. Samples of T i 0 2 (1.5 mm in thickness) were cut from a boule of the undoped single crystal (rutile) by use of a diamond saw. The surface tu be used was the (100). The ‘To whom correspondence should be addressed. F’rtsent address: Faculty of Integrated Arts and Sciences, Hiroshima University, Naka-ku, Hiroshima 730, Japan.

crystal slice was mechanically polished with a fine emery paper and then sealed in a quartz tube with 100 Torr of Br2 and 0.1 g of N b for N b doping onto the TiOz by vapor-phase transport at 800 “ C for 3 h. The sample thus doped was then removed from the tube, repolished mechanically, and etched in molten NaOH at 450 O C for 1 h. Finally, the crystal was rinsed in distilled water and then kept in boiling 20 N nitric acid for 1 h to remove a white layer of NaOH from the crystal surface. Ohmic contact was made by Ga-In alloy. The n-TiO, samples thus prepared ranged in ~ , estimated donor concentration from lozoto 0.5 X 1019~ m - when from a slope of a Mott-Schottky plot with a dielectric constant of 100. Therefore, they were heavily doped to match a critical condition for donor concentration where the electrode quantum efficiency reaches a m a ~ i m u m . ~ * ~ Photoelectrolysis Cell. The n-TiO, electrode was mounted in photoelectrolysis cell constructed of Pyrex glass and brass. The cell consisted of two compartments separated by a fritted glass: a smaller compartment is for a Pt counterelectrode and the other for photooxidation of hydrocarbons at the n-Ti02. The area of T i 0 2 exposed to the solution was 0.28 cm2, defined by a hole of a thin Teflon washer. The cell was designed so as to eliminate or minimize a contact between the acid and materials other than glass and TiO, because of the highly corrosive nature of the acid solution a t high temperature^.^ The Ti02 sample sitting at a bottom of the cell was illuminated with a 100-W high-pressure Hg lamp through a Pyrex tube with a flat bottom. The optical path length through the solution was ca. 15 mm. Potentiostatic measurements were made with Princeton Applied Research potentiostat/galvanostat (Model 173) by using a Ag wire as the reference electrode. The Ag wire was immersed into a main compartment of the cell. In some cases photocurrent measurements were made by using a two-electrode system without the Ag reference electrode to check the effect of corrosion of Ag on experimental results. However, no appreciable effect was observed on the behavior described below. All of the measurements were carried out in a thermostatic bath (Haake, Model FT)filled with ethylene glycol, which was maintained at a given temperature around 100 O C within an accuracy of f0.5 “C. Surface Analysis. Auger analysis of the n-Ti02 surface was performed on a PHI 595 Scanning Auger microprobe (Physical Electronics) instrument. An IS1 DS-130 scanning electron mi( 1 ) Harima, Y.; Morrison, S. R. J . Electroanal. Chem. 1987, 220, 173. (2) Gramstad, T. Haszeldine, R. N. J . Chem. SOC.1957, 4069. (3) Tamura, H.; Yoneyama, H.; Iwakura, C.; Sakamoto, H. J. Electroaml. Chem. 1977,80, 357. (4) Gautron, J.; Lemasson, P. L.; Marucco, J. F. Faraday Discuss. Chem. SOC.1980, 70, 8 1. ( 5 ) Sarada. T.; Granata. R. D.; Foley, R.T.J. Electrochem. SOC.1978, 125, 1899.

0022-3654/88/2092-5716$01.50/00 1988 American Chemical Society

Anodic Processes on n-Ti02

Figure 1. Cyclic voltammograms on a Pt electrode in triflic acid monohydrate at 130 OC under (a) and (b) Ar atmosphere and under (c) propane atmosphere. Potential-sweep rate: 180 mV s-I.

croscope equipped with an EG&G Ortec X-ray energy-dispersive system was used to examine the surface structure and to carry out the elemental analysis.

Results and Discussion Optical Properties of Triflic Acid Monohydrate. As is shown later, observed photocurrents sometimes decayed with time. In this section we describe the optical absorption characteristics of the solvent that were measured to check whether the solution is responsible. Without agitation by gas flow, the solvent kept at a high temperature remained transparent in the visible and U V region for many days. However, the solvent changed its color to pale brown after bubbling gas for many hours, as is reported in a literaturee6 The solution showed a small absorption peak at 450 nm (normally, optical density 4 V) to allow breakdown or tunneling to occur. Reversibility of the photocurrents is seen in Figure 4, where the photocurrent on n-Ti02 biased at 2.0 V is recorded versus time while the purged gas is repeatedly switched from Ar to propane and back. The delay in the response in the photocurrent after the initiation of propane purging can be explained as due to the

Harima and Morrison

5718 The Journal of Physical Chemistry, Vol. 92, No. 20, 1988 N

;

Ar

1

I

I

u 20E A,

+.

; .

a c

10-

0 - -

0

30

20

10

40

50

60

Tlmeimin

Figure 4. Photocurrent-time curve on n-Ti02biased at 2.0 V (versus Ag) in triflic acid monohydrate at 100 OC while purging propane or Ar. 1.7}

Figure 6. Plot of I h / l e versus P(C,H,) for propane oxidation on n-Ti02 in triflic acid monohydrate at 100 "C, where 1, and I, denote hole current

and electron current, respectively.

0

02

04

0.6

08

1

C P(C3Hd/atm

Figure 5. Change of current-doubling factor for propane oxidation on n-TiO, with varying partial pressure of propane mixed with (a) Ar and (b) 0, while keeping a total pressure constant at 1 atm; temperature 100

C)

* c

v

0.5 mA crn-2

"C.

E s c

time required for the gas to reach the Ti02surface by convection and/or diffusion. In fact, similar experiments carried out at various temperatures ranging from 50 to 135 O C showed that with increasing temperature the delay time tended to reduce, most likely due to a decreasing viscosity of the solvent, which can enhance mass transfer of propane in the solution. It should be noted here that the ratio of photocurrent with propane to that with Ar (current doubling, CD, factor) did not depend on the temperature above 70 "C but remained constant. Temperatures lower than 70 OC led to a smaller CD factor with increasing delay time, and C and below. The no effect of propane was observable at 60 ' lack of dependence of the photocurrent ratio on the temperature above 70 OC where no serious viscosity effect exists is not surprising if we ascribe the increase in photocurrent to a current-doubling mechanism. A current-doubling factor for excess amount of a current doubler is determined by the ratio of rate constants for two competitive reactions: one is an electron injection process from an oxidized intermediate into an semiconductor and the other a quenching of the intermediate by solvent or electrolyte. By this reason, the temperature dependence of the two reactions could tend to cancel, giving a constant current doubling. Because the photogeneration of holes is the rate-determining step for the background reaction (Figure 2), it is presumably not possible that the increase in saturated photocurrent shown in Figures 3 and 4 can be ascribed to an additional flow of photogenerated holes to the electrolyte. Figure 5 shows dependences of propane partial pressure, P(C,H,) on current doubling under the weak illumination, where propane mixed with (a) Ar or (b) O2with varying ratios of their flow rates was purged into the solution while keeping the total flow rate constant. In Figure 5a for the Ar and propane mixture, the CD factor is seen to increase abruptly with the increase in P(C3H8)and almost saturates at P(C3H8)values beyond 0.4 atm. The use of O2in place of Ar resulted in less CD factor in the whole P(C3H8)range. The reduction in CD factor for propane purged with O2can be interpreted simply in terms of a quenching (oxidation) of a highly reducing intermediate, responsible for the excess photocurrent, by 02.We notice, furthermore, that photocurrents for Ar alone coincide with those for O2 purging under the same experimental conditions. This finding may suggest that the solvent oxidation itself is not a current-doubling reaction, for if it were, the bubbling with oxygen would remove the intermediate, lowering the current for this case also. According to a

0 *

I

0

f 0

0

I

I

I

0

10

20

Time/min.

Figure 7. Effect of light intensity on current doubling during propane oxidation on n-TiO, biased at 2.0 V in triflic acid monohydrate at 100 'C. The light intensity is increased in the order of a-d. Up arrow and down arrow indicate, respectively, the time when purged gas is switched from Ar to propane and vice versa.

normal analysis of current doubling,8 the data of Figure 5a are replotted in Figure 6 as Ih/Ie versus P(C3H8)-', where Ih and I, are hole current and electron current, respectively. The plot yields a straight line with 0.55 atm for slope and 1.8 for intercept, suggesting that the current-doubling reaction of propane consists of a simple mechanism under the weak intensity of light. Photocurrent Decay during Propane Oxidation under Stronger Illumination. Figure 7 depicts a general trend of the effect of light intensity on the photocurrent-time curve where the purged gas is switched from Ar to propane and back. Under the weakest illumination (curve a) photocurrents remain constant over a long period of time of propane purging, as is seen also in Figure 4. Under stronger intensity of light (curves b and c), however, photocurrents during propane purging show a considerable decay with time. In some experimental runs, similar photocurrent decays were observed even during Ar purging, but in this case the lower currents were temporary. A sharp current spike would appear and then the photocurrent would recover to the original value. On the basis of the visual observation that the photocurrent spikes were synchronized with a release of bubbles from the electrode surface, this pattern of decay was considered trivial, ascribed to (8) Hykaway, N.; Sears, W. M.; Morisaki, H.; Morrison, S. R.J. Phys. Chem. 1986, 90, 6663.

Anodic Processes on n-TiOz

2-

N

1-

'E, 4

E

0-

, * =; 2 a

:

1-

0

a

(L

0-

Figure 8. Characteristics of photocurrent decays during propane oxidation on n-Ti02 biased at 2 V.

the growing of bubbles on the Ti02 surface leading to a scattering of light. Obviously, the trivial cause for the decay does not apply to the observation in curves b and c. It should be noticed in Figure 7 that the photocurrent values before and after propane purging are almost the same in spite of the decrease in the propane-induced current, although longer illumination under propane led to a slightly smaller photocurrent. The lack of change in the photocurrent when there is no current doubling (no propane) shows clearly that the photocurrent decrease under propane is associated with a decrease in excess photocurrent, not a decrease in hole current. The pattern of curve b or c was found repeatable with intermittent propane purging, as in Figure 4. Decay characteristics are investigated further in Figure 8. When, with propane kept purging, the light was turned off at a certain moment during the decay and then was turned on later, photocurrent values just before turn-off and just after turn-on coincided irrespective of the duration of the dark period (Figure 8a). On the other hand, photocurrent values after reillumination coincided with the values before propane purging in the case that the purged gas was switched from propane to Ar simultaneously with turn-off of light (Figure 8b). The pattern of curve b in Figure 8 was repeatable. These findings imply the following: the photocurrent decay requires both illumination and propane purging; the surface condition of the n-TiO, is kept unchanged in the dark despite an agitation of the solution with propane, and the cause of the decay can be removed by illuminating the electrode under no propane. On these bases, it may be suggested that during propane oxidation some insoluble species accumulate on the n-Ti02 surface, which tend to prevent a current doubling by a reaction (perhaps reduction) with a highly reducing intermediate from propane and also can be removed from the electrode surface by being oxidized by photogenerated holes. Under the most intense light (curve d in Figure 7), the C D factor was the smallest and the decay of photocurrents under propane was not so rapid. In this case, however, the decay was caused mainly by a decrease in hole current; oxidation of propane for 2 h resulted in a 50% decrease in photocurrent. Without propane (solvent oxidation) there was no decrease in photocurrent for many hours. An important observation, made visually under intense light, was that the gas evolution from the electrode surface, observed under Ar purging, stopped completely when propane was bubbled into the solution, beginning again when Ar was restored. This finding means that the hole reaction is changed completely, from some reaction yielding a gaseous product to propane oxidation when propane is added. Further we can suggest that as only a small CD factor is observed, only a part of the hole current associated with propane oxidation contributes to current doubling under the strong illumination. After ca. 20 C cm-2 of photocurrent flow across the n-TiO, surfaces, depending on whether the purging was with Ar or propane, the surface of the electrode exhibited a clear difference. The electrode surface after solvent oxidation was etched and appeared white, an appearance that was most likely caused by a light scattering on the surface roughened due to photocorrosion. The electrode surface after propane oxidation had a white layer on top of less etched surface. The white layer,

The Journal of Physical Chemistry, Vol. 92, No. 20, 1988 5719 insoluble in most organic solvents and insulating as is shown below, is likely responsible for the decrease in hole current a t very high light intensity. SEM and AES Analysis of the White Product. A preliminary survey was performed by using a scanning electron microscope (SEM) combined with electron probe microanalysis. Before measurements, a TiO, sample with the white layer was washed thoroughly with hot distilled water for several hours to remove the acid solvent and then it was dried. When a low accelerating voltage (8 kV) was employed, a signal of Ti from substrate TiOz disappeared and the white layer was found to consist of fluorine and sulfur, both presumably originating from the solvent. Elements of small atomic numbers such as oxygen and carbon are beyond the detection range of the equipment. On the other hand, Auger electron spectroscopy (AES) analysis without sputtering failed to provide any useful information on surface composition because of a serious charging effect, indicating that the film on the TiOz is electrically highly insulating. However, by sputtering the surface with Ar' (50 nA, 1 mm2) for a few minutes, the measurement gave a clear AES pattern showing the presence of C (275 eV) and F (660 eV). The atomic ratio of C to F was estimated as ca. 4, irrespectively of a duration of sputtering. It should be noted that no peak due to S, the presence of which was evident by SEM, was observed in AES spectra. This together with no detection of 0 suggests that any SO,or S03Hgroup in the solvent molecule is removed away from the insulating film by Ar' sputtering. We are unable to estimate the composition of the film definitely on the basis of the atomic ratio of 4 for C to F, because elements other than S and 0 might be taken off by sputtering. However, there is little doubt that the f i produced during propane oxidation is a polymeric species comprising the solvent and probably propane (as propane was necessary for the film formation), in part. In addition, it is reasonable to assume that the observed decay of photocurrents under moderately intense light (Figure 7b,c) is associated with a formation of higher molecular weight compounds from propane. Current Doubling during Oxidatiort of Chemicals Other Than Propane. Oxidation reactions of methane, ethane, n-butane, n-hexane, and methanol were also investigated on the illuminated n-TiOz electrode in triflic acid monohydrate. Purging of ethane led to a slight increase of photocurrents (less than lo%), while that of methane had no effect on the photocurrent-voltage curves at temperatures up to 135 O C . Furthermore, no appreciable change in the rate of bubble evolution under intense light was observed when methane or ethane was purged into the solution. In the case of n-butane the CD factor was ca. 1.15 when measured under weak light. On these bases, we can assume that a relative ease for oxidation of these simple alkanes on the illuminated n-TiOz follows the order of propane > n-butane > ethane > methane. Methanol and n-hexane were introduced into the solution maintained at 100 O C by using an Ar gas as a carrier. Even when the n-TiO, was illuminated with a weak light, measurements with n-hexane gave a response similar to those in Figure 7b,c, which were obtained for propane under a moderately intense light. A CD factor at a peak was 1.4, and cycling back to Ar without n-hexane, photocurrents settled down to the original value before the introduction of n-hexane into the solution. Methanol, on the other hand, is well-known as a current doubler in aqueous and nonaqueous solvents. However, an immediate 10% increase of photocurrent after its addition was followed by a gradual increase over 30 min to 50%. This slow change of photocurrents may be ascribed to a slow mass transfer of the reagent. Oxidation Process on Heavily Doped Ti02 in the Dark. An attempt was made to oxidize alkanes by holes generated by a reverse breakdown of a heavily doped n-Ti0, (Nd= lozo~ m - ~ ) . It was anticipated that holes produced by breakdown could be equally as effective as photoproduced holes.9 Figure 9a shows a series of current-voltage curves obtained by scanning a potential of the n-TiOZ repeatedly, where anodic (9)Sears, W. M.; Morrison, S. R. J . Phys. Chem. 1985, 89, 3295.

Harima and Morrison

5720 The Journal of Physical Chemistry, Vol. 92, No. 20, 1986

I;>t E'v

EIV

"* A 9

Figure 9. Currentvoltage curves obtained (a) during initial potential scans and (b) after potential scans for I h by use of heavily doped n-TiO, in the dark,

currents are seen to increase with time. The I-Vcurve of Figure 9b was recorded after cycling the potential for more than 1 h. In most cases, the increase in anodic currents was readily attained by biasing the electrode at 5 V in the dark or, in some cases, by an intense irradiation of light onto the n-TiO, biased at 5 V. Alkane oxidation was attempted by using the electrode treated in this way. However, no effect of purging methane, ethane, or propane was observed on the I-Vcurve within a potential range up to 5 V. In addition to this, a gas evolution from the surface of the n-Ti0, biased at 5 V did not cease when either of the alkanes was intmluced. A surprising finding was that the electrode surface stayed clean and shiny with or without alkanes after a current flow of ca.3 mA cm-, (at 5 V) for 1 h. The above findings indicate that neither alkane oxidation nor T i 0 2 corrosion occurs to an appreciable rate on the reversebiased n-TiO, electrode in the dark. Mechanisms of Anodic Processes on Illuminated n-TiO, and Reverse-Biased n-TiO, in the Dark. We have studied alkane oxidation by photogenerated holes and also have attempted to promote alkane oxidation by breakdown-produced holes. In this section we will discuss a mechanism of alkane oxidation on illuminated n-TiO, electrode and a plausible reason for no apparent alkane oxidation on reverse-biased n-TiO, in triflic acid monohydrate. We do not intend to postulate specific sets of reaction sequences but confine our attention to primary charge-transfer steps on the n-Ti0, under illumination and in the dark because the entire process appears to be much t m complicated to be analyzed. Electrochemical behavior of hydrocarbons has been intensively studied on noble-metal electrodes in strongly acidic media at elevated temperatures. The results on alkanes or saturated hydrocarbons that resist oxidation show that a dissociative adsorption precedes their oxidation.lw12 In triflic acid monohydrate used in our experiments, oxidation of some alkanes on Pt electrode is also known to be possible at less anodic potential than that of solvent oxidation, as is seen in Figure 2. It is quite reasonable to imagine that an alkane adsorption on Pt preceding their oxidation is prerequisite in this acid solvent as well. However, we feel that this reaction route is unlikely for alkane oxidation on the n-Ti02 semiconductor electrode. For one thing, adsorption activity of the oxide semiconductor should be weak compared with transition-metal electrodes. Second, even on the Pt electrode steady-state currents for alkane oxidation are small, in the range of 1-10 pA an", compared to the milliamp range observed here. The third reason is based on the linear relationship of I,/I.versus 1/P, suggesting no complicated behavior which could be expected if reactant adsorption were involved. For these reasons, we (IO) Niedrach. L. W.; Gilman. S.; Weinstock, 1. J. Eleelroehem. Soe. 1965, 112, 1161.

W. J. Electrochem. Soe. 1966. 113,645. (12) BNmmcr,S. E.;Ford, J. I.;Turner, M. I. J.Phys. Chcm. 1966.69, 3424. (11) Niedraeh. L.

"..li9

Figure 10. Band diagrams of (a) illuminated n-TiO, and (b) reversebiased n-TiO, in the dark, and density of states for solution species. S , P, P,and M denote solvent, propane, propane radical, and methane, respectively. State density of P is illustrated in such a way as having a large overlap with a conduction band edge of n-TiO,. Only the energy level, not the integrated density of states or the reorganization energy, is intended to have significance.

presume that the oxidation of alkanes by photogenerated holes is a process with no complication by reactant adsorption. Another possible reaction path that might be involved in alkane oxidation on the illuminated n-TiO, is an initial formation of a hydroxyl radical from a solvent oxidation followed by the reaction O H C3H, H,O + C3H,' The propane radical thus formed is an intermediate species that can inject an electron into a conduction band of n-TiO,, resulting in current doubling. In fact, Bockris et al. have demonstrated that on metal electrodes a water discharge precedes oxidation of unsaturated hydrocarbons such as If this is the case for our alkane oxidation on the illuminated TiO,, we expect it on the reverse-biased TiO, electrode as well because a gas evolution (presumably due to oxidation of H,O from the solvent leading to O H radical formation) was observed on the latter electrode in the dark in solutions without alkane. However, no sign of alkane oxidation was detected on the reverse-biased Ti0, in the dark. Because of this, we presume independent oxidation of solvent and alkane on the n-TiO, electrode. The schematic diagrams of Figure 10 suggest energy band diagrams for (a) the illuminated n-Ti0, and (b) the n-TiO, biased at a highly anodic potential in the dark, and its contacting electrolyte. State densities for reducing species in solution are described symmetrically in shape, as is originally predicted by Geri~cher'~J'and is evidenced by some experiment^.'^-'^ The positions of occupied states for alkanes are tentative. However, their relative order may be true in view of a close relationship between an ionization potential of a hydrocarbon and its oxidation p ~ t e n t i a l . ~ ' ~On ~ these bases, the observed preferential photooxidation of propaneover solvent oxidation can be interpreted most simply in terms of a direct isoenergetic transfer of photogenerated

+

-

1131 Bnkris. I. O.'M.: Wroblowa..H.:.Gilcadi.. E.:. Picrsma. 8.1. Famdav ChemSoe; 1965.41.2531. (14) Dahms. H.; Bockris, J. O M . J. Eleelroehem. Soc. 1964, 111,728. (IS) Johnson, J. W.; Wroblowa, H.; Bockris, I. O M . J. Electrochem. Soc. 1964 111. 863. ((6) Girischer. H. Z . Phys. Chcm. 1960, 26, 223 (17)Gerischer. H.Z . Phys. Chem. 1961, 27, 48. (18) Morriwn. S R Kktrochrmistr) ot Semironductorr and Oxidized Mela1 Elnirodrr; Plenum Ncu York. 1 Y 8 0 p 193 (19) Ysndrn Berghe. R. A. L.. Cirdon. t., Games. W . P. SwJ Sn.1973.

Di&&

~

2"

,'Q

_. _I"".I.

(20) Morrison. S. R. Surj. Sei. 1969. 15, 363. (21)Gleria, M.;Memming, R. Z . Phys. Chrm. 1976, 101, 171. (22) Pettinger, B.; khoppel, H.-R.; Gerischer, H. Ber. Bumcn-Ges.Phys. Chem. 1 1973.77.960, 9"' '' OLn (23) Pettinger, E.;Sehoppel. H.-R.; Gerischer, H. Be,. Bemen-Ges. Phys. (24) Neikam, W. C.; Dimeler. G.R.; Dcsmand, M.M.J. Electrochem. Soe. 1964,111, 1190.

(25)Fleischmann, M.; Pletcher. D. Telrohedron Lett. 1968,6255.

J. Phys. Chem. 1988,92, 5721-5726 holes from the valence-band edge to the reducing agent in solution as is shown in Figure loa. Without propane in solution, of course, the photogenerated holes will attack solvents directly. Smaller CD factors for alkanes other than propane can be explained by their lower hole-capture cross sections or smaller overlap of their fluctuating energy levels with T i 0 2 conduction-band edge. It seems to be of interest to compare the results in this acid solvent with those in fluorosulfonic acid (HS03F), one of the superacids. In the latter solvent, alkanes are known to be oxidized at a diffusion-controlled rate on Pt at less anodic potential than the potential where solvent oxidation start^.^^,^^ Therefore, the oxidation of alkanes in place of solvent oxidation is an exothermic reaction here. In addition, the oxidation is believed to involve the formation of an alkane radical which has a highly reducing ability. However, increase of photocurrents and any visually appreciable change were not found on the n-Ti02 before and after the solvent was purged with methane, ethane, or propane.28 This may imply no occurrence of an exothermic reaction in the superacid in accord with the result in triflic acid monohydrate. A final problem to be discussed is a cause for a sizable anodic current flow, not associated with alkane oxidation, on the re(26) Bertram, J.; Coleman, J. p.; Fleischmann, M.; Pletcher, D. J . Chem. SOC.,Perkin Trans. 2 1973, 394. (27) Coleman, J. P.; Pletcher, D. J . Electroanal. Chem. 1978, 87, 11 1. (28) Harima, Y.; Morrison, S. R., unpublished work.

5721

verse-biased n-TiOz electrode in the dark. Clearly, avalanche breakdown of the material or Zener breakdown can be ruled out because the breakdown-produced holes should have the same fate as photogenerated holes. Instead, one possible path is an electron tunneling from the solvent, through a narrow space charge layer, into the conduction band as indicated in Figure lob. A simple calculation using eq 2.22 in ref 18 with lozocm-3 for donor density and 100 for dielectric constant of the Ti02 leads to a value of ca. 50 A for the thickness of the space charge layer at 1.7 eV below the conduction-band edge when the electrode is biased a t 4 V. The layer under the described condition appears to be somewhat thick for efficient electron tunneling to occur. It should be noted, however, that this estimate is made for the n-Ti02 before the electrode treatment. If imperfections or energy levels in the bandgap in the bulk are generated in some fashion by the electrochemical or photoelectrochemical treatment of the T i 0 2 in the strong acid at high temperatures, they may assist the electron transfer from a solvent molecule into the conduction band. The observed difference of corrosion behavior of the material can be explained by the proposed mechanisms for a charge transfer at the semiconductor-electrolyte interface on the illuminated n-TiOz (involving holes) and the reverse-biased n-Ti02 in the dark (no holes generated to cause corrosion). Registry No. TiO,, 13463-67-7; propane, 74-98-6; triflic acid, 149313-6; methane, 74-82-8; ethane, 74-84-0; n-butane, 106-97-8; n-hexane, 110-54-3; methanol, 67-56-1.

Nonlonic Rodllke Micelles in Dilute and Semidilute Solutions: Intermicellar Interaction and the Scaling Law Toyoko Imae Department of Chemistry, Faculty of Science, Nagoya University, Chikusa, Nagoya 464, Japan (Received: October 15, 1987; In Final Form: February 8, 1988)

The static and dynamic light scattering for aqueous NaCl solutions of heptaoxyethylene alkyl ethers (C,E7, n = 10, 12, 14, and 16) has been measured, and the characteristics of nonionic rodlike micelles have been analyzed at finite micelle concentrations, considering the micelle growth and the intermicellar interaction. While the second virial coefficient of rodlike micelles is small and scarcely depends on NaCl concentration and the alkyl chain length, the hydrodynamic virial coefficient is negative and the absolute values increase with an increase in micelle aggregation number, comparable to the increase of the friction coefficient. The solution behavior of rodlike micelles in dilute and semidilute regimes has been discussed on the basis of the scaling laws. The scaling laws obey to the relations with the characteristic exponent v = 0.58-0.71. The deviation from the scaling laws was observed at higher micelle concentrations, suggesting the transition from the semidilute regime to the concentrated regime.

Introduction The development of the laser light source and dynamic light scattering instrument has promoted the investigation of micelle size and shape. Especially, the simultaneous measurement of static and dynamic light scattering has been available to analyze the characteristics of large or rodlike micelles. Young et al.' and Appell et aL2v3have measured the angular dissymmetry and the autocorrelation function from laser light scattering. They employed both quantities in order to distinguish between the possible shapes of large micelles formed at finite micelle concentration^'-^ and estimated the flexibility of elongated rodlike micelles with the aid of the distinct experimental investigati~n.~.~

Candau et al.495have performed the intensity and correlation measurements for the light scattered from aqueous salt solutions of cationic surfactants within a concentration range from dilute to semidilute regimes. They have reported that the shape of micelles changes, depending on alkyl chain length and surfactant concentration," and have first mentioned that the long flexible micelles entangle in the semidilute regime, obeying the scaling

law^.^^^ The overlap or entanglement of rodlike micelles was first evidenced by Hoffmann et al. for aqueous solutions of tetradecylpyridinium n-heptanesulfonate6 and cetylpyridinium ~alicylate.~ (4) Candau, S. J.; Hirsh, E.; Zana, R. J . Phys. (Les Ulis, Fr.)1984, 45, 1263. (5) Candau, S . J.; Hirsh, E.; Zana, R. J . Colloid Interface Sci. 1985, 105, 521

(1) Young, C. Y.; Missel, P. J.; Mazer, N. A.; Benedeck, G. B.; Carey, M. C. J . Phys. Chem. 1978, 82, 1375. (2) Appell, J.; Porte, G. J . Colloid Znterface Sci. 1981, 81, 85. (3) Appell, J.; Porte, G.; Poggi, Y. J. Colloid Interface Sci. 1982, 87, 492.

0022-3654/88/2092-5721$01 .50/0

( 6 ) Hoffmann, H.; Rehage, H.; Platz, G.; Schorr, W.; Thurn, H.; Ulbricht, W. Colloid Polym. Sci. 1982, 260, 1042. (7) Hoffmann, H.; Platz, G.; Rehage, H.; Schorr, W. Adu. Colloid Znterface Sci. 1982, 17, 275.

0 1988 American Chemical Society