6146
J . Phys. Chem. 1984, 88, 6146-6152
solutions will be unfamiliar. The method is immediately applicable however in the case that H i s a general even function of J, as long as there are no near intersections of the multiple-level curves. In this case the use of (A.6) for both the eigenstates of h will give an approximation to two of the many eigenstates of H. Of course, this method fails if other level curves come too close to the pair which is nearest the 0 axis. If the Hamiltonian H a n d h have two turning points and the intervening region is classically allowed, then the respective mapping of both turning points specifies a single energy curve e(E) since the action
is a canonical invariant. If the region between the two turning points is classically forbidden, as in the barrier penetration problem, the energy curve €(E)is also specified since the imaginary part of the action
like @ ( E ) ,is a canonical invariant. (The proof depends on the fact that for a small complex closed curve the canonical transformation can be approximated by its Jacobean, which being linear
and real preserves both the real and the imaginary part of the action.) Finally, in the case where a phase curve comes close to the 0 axis without touching it, as in the case of forbidden reflections over a barrier, the energy curve € ( E )is specified by the invariant
If more turning points are to be fitted, then h(g,p) will have to have free parameters. The restriction of H to be symmetric in J is a severe one, but as was shown in section 2, for each energy one can easily make the relevant pair of level curves symmetric with respect to the 8 axis. In the corresponding quantal representation we obtain wave functions which are accurate near the turning points, where the Hamiltonian is nearly symmetric, and whose asymptotic form away from these regions coincide in phase but not in amplitude with the correct WKB-like wave functions. This is because the reflection symmetry of a pair of curves need not apply to its neighbors. Since the semiclassical approximation to the eigenenergies depends only the phases, there is no restriction to the use of the present uniform approximation for this purpose. For the calculation of accurate wave functions for a general Hamiltonian, the present method supplies only a transitional approximation from which WKB-like connection coefficients can be derived.
Heterogeneous Photocatalysis. Dynamics of Charge Transfer In Lithium-Doped Anatase-Based Catalyst Powders with Enhanced Water Photocleavage under Ultraviolet Irradiation J. Kiwi* and C. Morrison Institut de Chimie Physique, Ecole Polytechnique F5dPrale. CH- 1015 Lausanne, Switzerland (Received: June 18, 1984)
The addition of Li to Ti02-basedanatase enhances considerably the activity of this type of catalyst. Evidence is presented here for the influence of ion doping on the observed charge-transfer processes on the catalyst surface. It shows that Li doping promotes conduction band electron transfer. This effect is likely to contribute to the observed enhancement in hydrogen generation and oxygen photoadsorption with increased Li content. Electron microscopy and other structure studies show that Li diffuses into the anatase bulk, altering the structure of this material. Implications of such alterations in water photocleavage under UV light are discussed. Displacement of the conduction band to more positive values takes place upon Li doping of TiOz samples.
Introduction The search for energetically economic methods of water photocleavage is a topic of research in several laboratories working in the area of energy conversion processes.’ It has been known for many years2 that TiOz can accommodate massive defect levels, especially at the surface. The alteration of catalytic properties in a semiconductor by doping catalysts with ions of different valency was first observed by V e r ~ e y . Processes ~ leading to N 2 0 decomposition through Li-doped NiO catalyst have shown marked improvement of the catalyst due to Li d ~ p i n g . More ~ recently, N i 0 5 and Co304electrodes doped with Li have shown a better (1) (a) Kiwi, J.; Kalyanasundaram, K.; Griitzel, M. Struct. Bonding (Berlin) 1982, 49, 37. (b) Harriman, A., West, G., Eds. ‘Photogeneration of Hydrogen”;Academic Press: London, 1983. (c) Gratzel, M., Ed. “Sources York, 1983. (2) Barksdale, J. “Titanium”; Ronald Press: New York, 1966. (3) Verwey, E.; Haaijmann, P.; Romeijn, F.;Osterhout, G. Philips Res. Rep. 1950, 5, 173. (4) (a) Hauffe, K.; Schlosser, S. DECHEMA-Monogr. 1956,26,222. (b) Hauffe, K.: Grunewald. H.: Tranckler-Greese. R. 2. Elekfrochem. 1953, 57, 937. (c) Hauffe, K. Rev. Pure Appl. Chem. 1968, 18, 79. (d) Szabo, Z., Kallo, D., Eds. “Contact Catalysis”; Elsevier: New York, 1976. (5) Tseung, A.; Bevan, H. J . Mater. Sci. 1970, 5 , 604.
0022-3654/84/2088-6146$01SO/O
performance in oxygen reduction processes. Insofar as we are aware, there have been no studies made reporting the effect of Li doping on T i 0 2 and its effect on the photocatalytic properties of the material. Within this framework, we have begun a study on catalytic activities of dilute solid solutions of Li in TiO,. This work aims at providing a basis for the detection of effects due to the presence of Li in H2generation and 0, adsorption under UV light. Attention will be paid to the ionic solid state of the Li doping. Such a study may be able to provide a correlation between electronic and catalytic processes taking place at the surface. Experimental Section UV irradiations were carried out with a Rofin lamp. The Of irradiation flasks was 25 cm3* were degassed with Ar and the H2produced or O2 photoadsorbed was
on photolytic experiments have been reported previously.6 Catalyst loading by Pt via ion exchange frorn Pt(NH3)4(0H)2 was used. The same technique was employed as described in ref 6a. Samples of Li-doped Ti02 were prepared by Dr. Panek (Bayer (6) Kiwi, J.; Gratzel, M. J . Phys. Chem. 1984,88, 1302. (b) Yesodharan, E.; Gratzel, M. Helv. Chin. Acta 1983, 66, 2145.
0 1984 American Chemical Society
The Journal of Physical Chemistry, Vol. 88, No. 25, 1984 6147
Heterogeneous Photocatalysis TABLE I
Li(mixed), at. %
mixing temp, 'C
0.5
500 500 400 500 600 700 400 500 600 700 400
2.7
5.0 5.0 5.0 5.0 7.0 7.0 7.0 7.0 10.0 10.0 10.0 10.0
500 600 700
Li, wt % Ti% g
LiOH, g
mixed
found
Li(found), at. 56
199.7 198.4 100.82 95.71 88.89 82.64 78.10 88.61 91.73 92.30 84.63 86.62 87.19 91.73
0.3 1.6 1.59 1.51 1.40 1.30 1.76 2.00 2.07 2.08 2.82 2.88 2.90 3.05
0.04 0.23 0.46 0.46 0.46 0.46 0.65 0.65 0.65 0.65 0.97 0.96 0.96 0.96
0.036 0.19 0.38 0.38 0.38 0.40 0.54 0.57 0.58 0.59 0.84 0.87 0.87 0.87
0.46 2.13 4.13 4.13 4.13 4.34 5.81 6.13 6.24 6.35 8.6 9.0 9.0 9.0
AG, Krefeld, Uerdingen, West Germany). By a slurry technique, Li-doped T U 2 was prepared, similar to the one employed by Teichner et aL7 for the preparation of lithiated NiO. In order to produce Li doping within a reasonable time (3 h), an intimate mixture of LiOH and T i 0 2 (Bayer) was heated at temperatures between 400 and 700 "C. The actual Li content of the catalysts was analyzed by atomic absorption by Dr. Panek and results are shown in Table I. BET areas were measured in a Micromeritics 2205 instrument. Peroxotitanates shown in Figure 6 were determined by titration with IOu3 K M n 0 4 at p H 0. Details of this determination in which the liquid as well as the catalyst precipitate were agitated 4 h with permanganate have been reported before.6 Thermal-gravimetric analysis was performed on a Mettler 2000 instrument. Transmission electron microscopy was carried out with a Philips-300s instrument. The amplification of 400000 was attained at 100000 V to obtain the highest resolution in the electron microscope. Mobilities were measured on the Li-doped anatase particles situated in the stationary planes of a Mark I1 microelectrophoresis apparatus (Rank Brothers, Cambridge, England). Determination of the conduction band position of Li-doped anatase powders was carried out following a technique recently reporteds for these kinds of measurements.
Results and Discussion Photocatalytic H2 Evolution on Li-Doped TiO,. Figure 1 presents the results of the H2evolution rate for UV irradiations of a platinized catalyst of TiO, (Bayer) containing different amounts of Li. In all cases a flux of 170 mW/cm2 of a 150-W Xe lamp was used; 0.05% Pt loading on T i 0 2 was employed in all cases. It provided the highest rate of H2evolution as compared with Pt loading carried out on TiO, powder by other techniques, e.g., photoplatinization, impregnation, and impregnation-calcination followed by reduction. A similar preparation by ion exchange loaded with 0.05% Pt showed the highest efficiency and 100% dispersion for the Pt clusters on Ti02.9 These results confirm the trend for the efficiencies in H2evolution on Ti02(P25 Degussa)? These observations show that the technique employed in the deposition of Pt on TiOz powders is the controlling factor in H, photoevolution and not the different surfaces of diverse Ti02 materials. In the present experiments it is readily seen that Li doping affects the rate of hydrogen generation and that the sample 0.05%Pt/Ti02-7% Li represents the most favorable situation for H, evolution. Different loadings of Li have been prepared by heating LiOH and Ti02 for 15 h at 200 "C and for 2-3 h at 500 "C. The 200 O C preparation temperature is relative low, below -0.3 of the Tammann temperature for ?'io2. From this observation it follows that the mobility of the Li ions at 200 "C inside the bulk of Ti02 is small. It is difficult then to imagine that (7) Teichner, S. Adv. Carol. 1969, 20, 107. (8) Ward, M.; White, J.; Bard, A. J . Am. (:hem. SOC.1983, 105, 27. (9) Ichou, I.; Formenti, M ; Teichner, S. In "Spillover of Adsorbed Species ; Pajonk, G., Teichner, S., Gerrnain, J., Eds.; Elsevier: Amsterdam, 1983; p 63. (b) Short, D.; Monsour, A,; Cook, J.; Sayers, D.; Katzer, J. J . Catal. 1983, 82, 299.
p)
500)
I
400
hours
-L.
Figure 1. Hydrogen evolution induced by UV light in 1 N NaOH solutions using 40 mg of catalyst 0.05% Pt/TiO, (Bayer) with an added amount of atomic Li at 500 O C : (a) 0%, (b) 0.5%,( c ) 2.7%, (d) 5%, (e) lo%, (f) 7%.
substitutional incorporation of Li could occur at 200 OC. Therefore, it is unlikely that bulk Li doping has taken place. Li ion doping will take place on the surface of TiO,. In work related to Li doping of N i 0 4 it has been observed that Li+ ions begin to enter the T i 0 2 lattice at -400 "C. The fact that Li+ does not have the same valence state as Ti4+ means that the samples prepared at 500 "C (with consequent remarkable increase in the observed H2 yields) involve an extra charge introduced by these ions. TiO, is an n-type excess semicond~ctor.~ When Li" is added to TiO,, an extra charge is introduced in the system due to this acceptor. Using the Kroger-Vink notation, one may express the defects introduced into TiO, in the following way:
+
Oo = '/202 V i '
+ 2n
(1)
or where 0, = oxygen ions on normal lattice sites, V{ = oxygen anion vacancy doubling positive charged, n = electrons in lattice, TiT, = Ti metal on a normal lattice position, TiT( = Ti metal on a normal lattice position with a negative charge. Since Li-TiO,
6148 The Journal of Physical Chemistry, Vol. 88, No. 25, 1984
catalysts were prepared in air, the doped TiOz will equilibrate with the air atmosphere. This means that the incorporation of Li as Li,O proceeds without adding extra oxygen from the gas phase in the equilibrium state. If L i 2 0 is incorporated into TiO, as acceptor, this acceptor will be compensated at low oxygen activities by oxygen vacancies. Balancing the equation for Li incorporation with respect to charge mass and lattice sites LizO = 2LiTjlll
+ Oo + 3V;
OC
200
400
I
I
Kiwi and Morrison
-
800
600 I
I
a
(3)
From eq 1-3 it follows that the acceptor doping through Li has resulted in the introduction of a sizable concentration of oxygen vacancies. These oxygen vacancies would be in direct proportion to the Li concentration used. This increase in oxygen vacancy concentration is consistent with the experimental findings reported in Figure 1 up to a level of 7% in Li doping. When UV light is absorbed by TiO,, an electron and a hole are formed as shown by eq 4. hv (>E,)
+ TiO,
-
eCB-+ h+
(4)
Li doping seems to facilitate the separation of photoelectrons and holes at the barrier.3,4110 This is shown in Figure 1. As all irradiations in Figure 1 have been carried out in 1 N NaOH, it is likely that the excess of OH- basic groups also contributes to the creation and maintenance of a Schottky barrier due to the negatively charged OH- groups on the surface.” This argument is not incompatible with an increase in tunneling efficiency which takes place at basic pH values. It is well-known that doping of Ti0, creates a narrower depletion layer.1° Band bending would then take place on a depletion layer with a narrower width and becomes more drastic. The increased band bending would allow electron tunneling to proceed more easily at the interface,” enhancing the rates for electron transfer. Since the 0.05% Pt/ Ti02-10% Li sample was less efficient than the 0.05% Pt/ TiO2-7% Li (Figure l ) , it is possible that 10% doping causes a decrease in the negative potential of TiO,. This decrease would be related to the corresponding decrease in the quantum yield of the photocatalytic reduction reaction. Such arguments have already been used in Li-doped surfaces involved in photocatalytic processes. 12-15 Table I presents the results for the atomic percentage of Li found vs. the amount of Li added during the preparation. The preparation process at 500 O C is described in the Experimental Section. From such a linear correlation, it is readily confirmed that bulk doping has taken place in the lattice. Figure 2 presents results for H2evolution when 40 mg of 0.05% Pt/Ti02-7% Li is irradiated in 1 N N a O H solution. Samples of the catalyst have been prepared at the temperatures shown in this figure. Since doping was carried out between 0.3 and 0.5 of the fusion temperature of TiO,, Li ions have enough mobility to penetrate quite homogeneously in the bulk and at the surBut heating at these relatively high temperatures produces a decrease in the BET area, sintering the samples. Since the bulk diffusion of Ti ions controls the surface area of these catalysts, the sintering process of the host species is reflected in reduced BET areas. Undoped TiO, (Bayer) was determined to be 180-200 m2/g. Determining the BET area for samples doped at 400-700 OC,and dividing the H2 yields at each temperature (10) (a) Kofstad, P. “Non-Stoichiometry, Diffusion and Electrical Conductivity in Binary Metal Oxides”; Wiley: New York, 1972. (b) Blickley, R.; Stone, F. Trans. Faraday SOC.1968, 64, 3393 and references therein. (11) (a) Wagner, F.; Somorjai, G. J . Am. Chem. SOC.1980, 102, 5494. (b) Ferrer, S.; Somorjai, G.Surf. Sci. 1980, 94, 41. (12) Zakharenko, V.; Cherkashin, A,; Keier, N. Kosheheev, S. Kinet. Katal. 1975, 16, 142. (1 3) Shobaky, G.;Gravelle, P.; Teichner, S. Bull. SOC.Chirn. Fr. 1967, 3246. (14) Shanon, R. Acta Crystallogr., Sect. A 1976, 32, 751. (15) “CRC Handbook of Chemistry and Physics”; Chemical Rubber Publishing Co.: Cleveland, OH, 1978. (16) Goldstein, J.; Tseung, A. J . Phys. Chem. 1972, 76, 3646. (17) Blickley, R.; Stone, F. In “Symposium on Electronic Phenomena in Chemisorption and Catalysis on Semiconductors, July 1968, Moscow”; Hauffe, K., Wolkenstein, Th., Ed.; de Gruyter: West Berlin, 1969; p 138.
b
I
I
I
200
400
600 O C
1
aoo
-c
Figure 2. (a) Media rate per hour of hydrogen produced under UV irradiation when 40 mg of 0.05% Pt/Ti02-7% Li catalyst is irradiated in 1 N NaOH. Temperature of diffusion of lithium into TiO, is shown on the abscissa. (b) Same as in part a-hydrogen yields as a function of m2 BET area of the Li-Ti02 catalyst prepared a t different tempera-
tures.
I
/
/
0
24
24
/
A
J
J
24
hours-
Figure 3. UV irradiation of 0.05% Pt/Ti02-7% Li catalyst; 35 mg in 25 cm3 of 1 N N a O H as a function of time of irradiation.
(Figure 2a) by the area available in 40 mg of catalyst, one arrives at the plot of H2 yields vs. temperature shown in Figure 2b. The maximum activity per milligram of catalyst in Figure 2a is found at 500 “C. But since the BET areas of the catalyst doped at 500 and 600 O C are 33 and 20 m2/g, respectively, the activity per square meter of catalyst at 600 “C is higher. This is shown in Figure 2b. Figure 3 shows repetitive cycles for H, formation when 0.05 Pt/TiO,-7% Li is irradiated several times. The rates of H2
The Journal of Physical Chemistry, Vol. 88, No. 25, 1984 6149
Heterogeneous Photocatalysis
I-'. t
t
v)
L
3 0
0"
-a-
b
C
a
b
e
a
b
4j
c 4300
100-
I L
I
I
I
0
2
6
10
I 14
hours-
Figure 4. O2 photoadsorption (left ordinate) and photoproduced H2 (right ordinate) when solutions containing 375 pL of O2are irradiated in 25 cm3 of degassed 1 N NaOH containing 40 mg of the following catalysts: (a) 0.05% Pt/Ti02, (b) 0.05% Pt/TiO,-0.5% Li, (c) 0.05%
Pt/Ti02-7% Li. The lines plotted in the figure are drawn through points taken at 30-min intervals in the O2 photoadsorption experiments. production in a 24-h period were reproducible up to 15%. Bubbling in the first three recyclings (as shown in Figure 3) will eliminate H2 and oxidative products formed during irradiation. The latter type of product will be dealt with in Figure 5 . In 10 recyclings, 14 mL of H2 was obtained from the same catalyst over a 240-h irradiation period. Since 35 mg of 0.05 Pt/Ti02-7% Li was added (4.5 X lo4 mol) and a total of 6.3 X lo4 mol of H, was obtained during these cycles, the turnover number for H,production exceeds 1. Therefore, H2 formation is catalytic with respect to TiO, as reported previously in work from this laboratory.6 Studies on the Oxidative Products of Water Photocleavage. Figure 4 presents the kinetics of oxygen photoadsorption for (a) 0.05% Pt/TiO,, (b) 0.05% Pt/Ti02-0.5% Li, and (c) 0.05% Pt/Ti02-7% Li. Runs were carried out in alkaline solutions at pH 14; 375 pL of 0, was injected in the three runs. Results are shown on the left-hand side in Figure 4. On the right-hand side of this figure the concomitant hydrogen production is shown as a function of time. The observed rate for H2 production (samples a-c) was the same as the reported rates in Figures 1 and 3, once the initial oxygen added was consumed. From these oxygen adsorption experiments, it is readily seen that the 0, adsorption capacity of Ti0, is changed due to Li doping. This observation has been reported before on Li-NiO and Li-ZnO samples adsorbing oxygen from the gas phase.'*J9 The adsorbed fraction of oxygen increases (as shown in Figure 4) when the Li content in the sample is higher. The decreasing rate observed for the oxygen adsorption in the three samples under study is due to the boundary layer produced on O2adsorption which makes subsequent absorption slow. A Schottky bamer is formed at the surface due to oxygen chemisorption1° as shown by eq 5 , and will move ecB-
+ 0,
-
0,-
to more negative values during the 0, photoadsorption process. Such a process as shown by eq 5 would take place at high pH since 0,at lower pH is converted to HO,.. Increased formation of 0,- centers for oxygen photoadsorption has been confirmed by other investigations on Li-ZnO samples.20,21Figure 4 shows then (18) Shobaky, G.; Gravelle, P.; Teichner, S. Bull. SOC.Chim.Fr. 1968, 3251. (19) (a) Stone, F. Adu. Curd. 1962,13, 1. (b) Barry, I.; Stone, F. Proc. R. SOC.London, Ser. A 1960, 255, 124. (20) van Hove, H.; Bohrmann, D.; Luyckx, A. Surf. Sci. 1967, 7, 474. (21) (a) Lisachenko, A.; Vilesov, F. Kine?. Kutul. 1972, 13, 420. (h) Zakharenko, V.; Chersakin, A.; Keier, N.; Gerasimova, G. Kinet. Kurd. 1975, 16, 143.
1
r
Figure 5. Amount of hydrogen produced (column a) when 25 f solution is irradiated with 40 mg of 0.05 Pt/Ti02-7% Li catalyst at pH 14, 7, and 2 for 8 h. In column b, the amount of peroxytitanate in the liquid is shown in each case. In column c, the amount of peroxytitanate particles associated with the catalyst surface is shown in each case.
c
\ \
100 300 500 700 O C Figure 6. Variation of weight loss with temperature: (a) TiO, untreated; (b) TiOz treated, pH 14; (c) Ti02-7% Li treated, pH 14.
that doping TiO, with Li at 500 "C develops the capacity for oxygen chemisorption of these catalysts. This capacity is not negligible since up to one monolayer of O2on TiO, can be adsorbed in liquids. When the light is turned off, the O2stays irreversibly adsorbed, which is a characteristic of photoadsorption processes. Figure 5 presents the results of irradiations carried out for 8 h on 25-cm3solutions when 40 mg of catalyst 0.05% Pt/Ti02-7% Li is added in solution. As seen from this figure in column a, the pH plays an important role in the amount of hydrogen obtained. The oxidative product formation is shown in columns b and c. When solutions are centrifuged after irradiation and the supernatant is titrated with M KMn04 at pH 0, the optical absorption decreases at 525 nm. In a second step, the catalyst precipitate separated from the solution is also reacted with
6150
The Journal of Physical Chemistry, Vol. 88, No. 25. 1984
Kiwi and Morrison
M KMnO, (at pH 0) and shows that oxidative products are formed in the bulk. This is shown in column c for the different pHs under study. Recent ~ t u d i e s ~have ~ ~ " identified peroxide formation (O?-) hy EPR and by infrared (IR)spectracopy when TiO, is irradiated in the UV. These species were reported to have been retained on the TiO, surface in all cases. IR spectral hands hetween 800 and 900 cm-' appear on UV irradiation due to the various modes of vibrations of T i 4 in the anatase surface. Tbese species have been ascribed to surface peroxytitanate complexes. It is seen from Figure 5 that at pH 14 about 1.ZX mol of peroxytitanate is produced in the liquid phase and 0.6 X 1WSmol is retained on the surface of the catalyst powder. The yields of peroxides were about 4 times higher at pH 14 than at pH 7. Yields for H, formation are also about 3 times higher at pH 14 than at pH 7 and this observations suggests the possible mechanism for peroxytitanate formation as shown in eq 6. This mechanism is
consistent with the following expeiiinentdl observations. (a) OH basic groups on the surface are more abundant at basic pIf More intermediate peroxide and te are detected in basic media Since this interm precursor of the H,, it explains the decrease observed more acidic (h) Adsorption of H,O1 color In our expertments long-time irradiations (SO h) produced an intense yellow color but only in alkaline media. (c) No oxygen was detected in these runs, showihg that uptake of the intermediate peroxide may take place. (d) Since peroxytitanate originate from intermediates that are formed from mobile OH groups at the TiO, surface, the loss of available OH basic groups will lead to a more on step for the remaining OH groups 2s ve product formation at more acidic pHs will he therefore lowered consistent with the results shown in Figure 5 In the exDerimenta1 observations shown in Fteures 1-4. 170 onniW;cm'light intensity obrr the full rp&Truin ,,,ys uied. I n thir !;*-02--Ti3+-ii4+~-02-.,~~i4+.n,experiment (Figure 5 ) the lamp iilten,il) u d b I I X I iiiW '.'ill' This intensity allaus the comparison of H, )iuldh reported in Figure 5 with a catalyst0.05% Pt/TiO, (P?SI k g u a w ) used prcvi,nisl).bd The present cata1)rt s h o w ll.(lb'; c o n r e r h n cllicicnc! a i cum-0 -..._____..0pared wiih the previous catal),t a t p l i I 4 Scverthcloi. the perox)titanate yields u r i c ahmt 50% of tlic a i n o u n i pchiblc i n w2 + -Ti*+-rp-Ti4+/O:\ 0, -Ti++-02--++(6) relation to i h r hydrogrn piduccd I n thc 0 055 PI, 1 i 0 2 (P25 (p-peroxohitonate mmplex Degiisra) this value reached abouf 20'4. wrfoce titanium peroxide Slnictural and Elecrra.hmicul ~ ' h , ' r ~ ~ , t ~ , ro,f~I "irIJoprd i~,~ 7i0, Catdictr. Tigurr h presents the ml>ilt,ation action of !he
p"
p"-
I I
I
I
I
I
I
__ I
I
(22) Gonzales-El& A.; Munuera, G.; Soria, J. J . Chem. Soc., Farad@y Tram. I 1979, 75,748. (23) Munuera. G.;Nsvic, 1. 'Proceedings of the 4thNatiional Meeting on Adsorption, %villa. Sepi 1979". p 25.
(24) Bmnstra, A.; Mutsaers, C . J . Phys. Chem. 1976, 80, 1694. (25) Hauffc, K.; Raveling, H.; Rein. D,%. Ph.w. C'hcm. 1977. 104, 89.
The Journal of Physical Chemistry, Vol. 88, No. 25, 1984 6151
Heterogeneous Photocatalysis
Ti02host matrix on the hydration and hydroxyl groups as revealed by the thermal behavior of the sample. Thermogravimetry (TGA) and differential thermal analysis (DTA) have been carried out for TiOz and TiO2-7% Li samples and these findings will be discussed in connection with other physicochemical measurements in these systems. Trace a (Figure 6) shows a 12-15% loss for TiOz (Bayer) up to 200 OC. As reported earlier,26water loss occurs as a multistage process. The observed loss agrees well with the loss of 1 mol of water per mole of TiO,. Trace b shows the results of TiOZsamples which have been heated at 500 OC, rehydroxylated and rehydrated in basic media (1 N NaOH, pH 14), and dried for 3 h under vacuum at 60 OC and subsequently for 2 h at 500 OC. Any reported weight loss for this “treated” sample will, therefore, reflect dehydration or OH- desorbed from the sample surface, since other constituents and impurities have been eliminated. Trace b shows dehydration taking place between 150 and 200 OC. This temperature is higher than the temperature necessary to promote the same process shown in trace a. Trace c shows only a small loss in weight for the TiO2-7% Li surface “treated” in the same way at pH 14. Stabilization action of the matrix on the hydrated and hydroxylated surface is therefore taking place. Since surface dehydration is largely eliminated in this case (as shown in trace c) Li doping may be involved in the suppression of surface dehydration.,’ Since in strong basic media the water on the surface is substituted by basic O H groups,28the process being hindered here is dehydration involving bimolecular surface interaction between neighboring hydroxyl g r o ~ p s .Li ~~,~~ ions may affect basic O H groups on the surface, decreasing drastically their mobility and, therefore, their possibility to interact with the few remaining H+ groups or basic OH, causing water loss. A small loss in weight at temperatures >200 OC is due to Figure 8. Electrophoretic mobility as a function of pH for lom2N constitutional water as well as O H groups. In traces a and b these (NaOH + HC1) solutions of (a) TiOz (Bayer), (b) Tio2-2.7% Li, (c) losses amounted to 3-4% up to 700 O C . Experiments were carried Ti02-7% Li, (d) Ti02-10% Li. out when the surface had been treated at pH 7 and 2. The behavior of these powders in the H2 generation processes was quite agglomeration into big unit cells when slurry TiO, is heated in different, proving the determining role of hydration and hythe presence of Li ions. droxylation. DTA analysis reveals an exothermic type of behavior Figure 8 presents the results obtained for electrophoretic mofor TiO,, Ti0,-7% Li untreated, and TiO2-7% treated. Such a bilities of the following samples: (a) TiOz (Bayer), (b) TiOz-2.7% behavior may correspond to the exothermic transformation of Li, (c) TiO2-7% Li and, (d) Ti02-10% Li. A solution of anatase to rutile since Li has been reported previously as a ruM ionic strength made of appropriate concentrations of NaOH tilization In the present work no DTA peaks were and HC1 was used in all electrophoretic measurements. This allows observed until 700 O C , which could be attributed to the intera constant width for the diffuse layer for all the particles under conversion of anatase to rutile. Therefore, the DTA measurements observation. From Figure 8 it is readily seen that a shift takes have been omitted for clarity of presentation in Figure 6. place in the isoelectric point (IEP) of anatase as Li doping inFigure 7a presents the results of electron microscopy studies creases from 5.4(Ti0, (Bayer)) to 6.3 (TiO2-7% Li). Since the on TiO, (Bayer). Particle size is seen to range from 300 to 1000 anatase used in this work was obtained by low-temperature hy8, since polyhedric crystals are aggregated into larger particles. drolysis from titanyl sulfate, sulfate ions cause the hydrolyzed TiO, The lower left of Figure 7a shows that the TiO, atomic planes to show an IEP at a more acid pH than TiO, (P25 Degussa) (IEP responsible for the porous structure are aligned in each unit. 6.6). Thus, some adsorption of sulfate takes place. Higher values Figure 7b shows a 0.05% Pt/TiO, (Bayer) catalyst. Since the for the IEP in the Li-doped samples indicate that positive charges catalyst was prepared by reducing the platinum at 400 OC under carried by the Li are determining in the observed mobilities around Hz, Figure 7b shows that the spongelike structure with atomic the IEP. Electrophoretic mobilities are therefore sensitive in the planes aligned is conserved in spite of the high-temperature region where the observed mobilities for species tend to zero. The treatment; 10-8, Pt particles are observed on the anatase, conspecies under consideration are TiOH,’ and TiO-. Electron firming previous This Pto cluster size for the same microscopy studies (Figure 7) show that particles of Li-doped TiOl loading has been observed in Ti0, (P25 Degussa). Figure 7c also catalyst aggregate in solution. Care was taken to follow the shows TiOz (Bayer)-7% Li in which the doping was carried out mobility of the smaller aggregates during electrophoresis. They at 500 OC. One observes that the crystallographic planes of TiOz consist of smaller particulate aggregates per kinetic unit and better are aligned in each of the smaller crystallites produced and not reflect the electric field interaction on the particles under obin each original unit of undoped TiOz as in Figure 7a. The T i 0 2 servation. (Figure 7c) shows a dense structure but the agglomerates are still Conduction band determination as shown in Figure 9 was between 300 and 1000 8, as already shown in Figure 7a. Figure carried out according to an electrochemical methods of collecting 7d shows that Pt loading on Ti0,-7% Li produces 10-8,Pt deposits the photogenerated charge on a Pt flag electrode immersed in the similar to the clusters seen in Figure 7b. The present electron irradiated suspensions. The solution used had a composition as microscopy experiments show the first evidence that Li diffusing described in the legend to Figure 9. Our reaction vessel was of into Ti0, alters the bulk structure of this material, avoiding the same design and volume as employed in ref 8. The voltage variation induced under irradiation was followed in a recorder (26) Boehm, H. Discuss. Faraday SOC.1971, 52, 264. which was fit to register current variation by means of an ap(27) Blickley, R.; Jayanty, R. Faraday Discuss. Chem. SOC.1974,58, 194. propriate resistance. Typical data from such experiments are (28) Gonzales-Elipe, A.; Munuera, G.; Soria, J. J . Chem. Soc., Faraday shown in Figure 9. Since the redox potential of the MVZ+/MV+ Trans. 1 , 1980, 76, 1535. couple is pH independent, in traces a-c, the observation of the (29) Kelly, P.; Braunlich, P. Phys. Rev. B 1970, I , 1581. (30) Bickley, R. Chem. Phys. Solids 1978, 7, 118. pH effect is related to changes within and/or on the surface of ~
~
~~~~
J . Phys. Chem. 1984, 88, 6152-6157
6152
(Bayer)), b (Ti02 (Bayer)-2.7% Li), and c (Ti02 (Bayer)-7% Li), the pHo values found were 6.5, 7.5 and 9.0, respectively. From the pHo values obtained in eq 7, and with the knowledge that the MV2+/MV+ redox potential is -0.69 V vs. SCE, Ef values of -0.23, -0.19, and -0.13 V are obtained for the compounds shown in traces a-c. Li doping shifts pHo in eq 7 to more basic pH values.
2ol
Figure 9. Dependence of rate of the change of photocurrent with time (AilAt) on pH for stirred and N2-purgedTiOzsuspension photocell. Cell conditions: TiOzpowder (250 mg); H,O (100 mL); [NaOAc] = 1.0 M; [KN03] = 0.1 M; [MVz+]= 1 mM. Platinum collector electrode at -0.20 V vs. SCE: (a) Ti02 (Bayer), (b) TiO, (Bayer)-2.7% Li, (c) TiO,
(Bayer)-7% Li. the Li-doped semiconductor powders used. The source of this effect is the shift of the Fermi energy level Ef with pH as shown by eq 7. Extrapolation of the two lines Ai/At when large and Ef = Ef(pH 0) - 0.059pH (at 25 "C) (7) small Ai/At changes were observed afforded a precision of fO.l pH unit in the determination of pH,. The intersection point (pH,) represents the pH value in eq 7 when the Fermi level (Ef) equals the redox potential of the couple MV2+/MVC. For traces a (Ti02
Conclusion In conclusion, an active Li-doped anatase catalyst in water photocleavage has been presented in this study. The first evidence has been presented here in which Li doping modifies the dynamics of charge transfer across the anatase interface, improving its water photocleavage efficiency. The link between the nature of the catalysis and the role of Li ions has been presented by using diverse physicochemical techniques. We think that the effect of doping on the electronic transfer in semiconductor powders can, to a considerable degree, be explained by the phenomena described in this paper. The role of surface hydroxyl groups in H, generation and O2photoadsorption is of great importance in photocatalytic reactions as confirmed by pH dependency of the products observed. This observation has been previously reported by Somorjai.1',31 Evidence has been presented which shows that modification of the semiconductor bulk by Li doping also has a profound effect on the photocatalytic products observed. The detailed mechanism of water photocleavage remains to be identified.
Acknowledgment. This study was supported by the Swiss National Science Foundation. We thank Dr. Panek of Bayer AG for the preparation of the catalyst samples and M. Buffat for his assistance with the electron microscopy work. Helpful discussions with Professor M. Gratzel are appreciated. C.M. thanks the European Photochemical Association for a travel grant. Registry No. H2, 1333-74-0;Li, 7439-93-2;water, 7732-18-5;anatase, 1317-70-0. (31) Ferrer, S.; Somorjai, G. J. Phys. Chem. 1981, 85, 1464.
Diffusion-Controlled Reactions between a Spherical Target and Dumbell Dimer by Brownian Dynamics Slmulation S. A. Allison,*+ N. Srinivasan, J. A. McCammon,* Department of Chemistry, University of Houston, University Park, Houston, Texas 77004
and S . H. Northrup Department of Chemistry, Tennessee Technological University, Cookeville. Tennessee 38505 (Received: June 18, 1984)
A new Brownian dynamics trajectory approach used recently to study the diffusion-controlled reaction of spherical reactants is extended to the simplest case of structured reactants: dumbell dimers reacting with a spherical target. It is shown that, for dimers with a single reactive subunit, electrostatic torques exerted on the dimer by the target can increase the reaction rate by "steering" the dimer toward productive collision geometries. The effects of variations in the reactive surface of the dimer and in the Coulombic and hydrodynamic interactions between the reactants are also considered.
1. Introduction since the pioneering work of Smo~uchowskiand Debye, who investigated the problem of diffusion-controlled reactions in the absence' and presence2 of centrosymmetric Coulombic forces, there has been a proliferation of theoretical studies of such reactions based on more complex models. These have included the effects +Current address: Department of Chemistry, Georgia State University, University Plaza, Atlanta, GA 30303.
0022-3654/84/2088-6152$01.50/0
of hydrodynamic interaction's4 and solvent caging effects.5 The general area of enzyme-catalyzed bimolecular reactions has received special attention. The rate of diffusional encounter of an enzyme and ligand may be influenced by many factors, such as (1) Smoluchowski, M. V. Phys. Z . 1916, 17, 557. (2) Debye, P. Trans. Electrochem. SOC.1942, 82, 265. (3) Friedman, H. L. J . Phys. Chem. 1966, 70, 3931. (4) Deutch, J. M.; Felderhof, B. U. J . Chem. Phys. 1973, 59, 1669. (5) Northrup, S.H.; Hynes, J. T. J . Chem. Phys. 1979, 71, 871.
0 1984 American Chemical Society