Photoinduced Adsorption of Hydrogen and Methane on γ-Alumina

In the first, emission originates with electron trapping by such deep energy traps as anion vacancies {e- + Va → F+ + hν1} to yield electron F-type...
0 downloads 0 Views 151KB Size
Langmuir 2004, 20, 129-135

129

Photoinduced Adsorption of Hydrogen and Methane on γ-Alumina. The Photoinduced Chesorluminescence (PhICL) Effect N. S. Andreev,† A. V. Emeline,‡ S. V. Polikhova,† V. K. Ryabchuk,† and N. Serpone*,‡,§ Department of Photonics, Institute of Physics, University of St. Petersburg, St. Petersburg, Russia, Department of Chemistry and Biochemistry, Concordia University, 1455 deMaisonneuve Boulevard West, Montreal, Quebec, Canada H3G 1M8, and Dipartimento di Chimica Organica, Universita di Pavia, Via Taramelli 10, Pavia 27100, Italy Received June 3, 2003. In Final Form: September 18, 2003 Adsorption of hydrogen and methane on a preirradiated surface of γ-Al2O3 produces an afterglow, which has been described as a photoinduced chesorluminescence (PhICL), whose spectral features identify with the intrinsic photoluminescence of alumina. The emission spectrum consists of at least four overlapping single emission bands. For methane adsorption, the PhICL phenomenon is seen only if the solid is preirradiated in the presence of oxygen. Emission decay kinetics of the PhICL effect for γ-Al2O3 reveal two wavelength regimes: a short wavelength regime at λ ) 300-370 nm (decay time τ ) 1.1 ( 0.2 s; signal width ) 2.8 s), and a longer wavelength regime at λ ) 380-700 nm (decay time τ ) 2.1 ( 0.1 s; signal width ) 4.3 s). A model is proposed in which there exist two different emission centers and, thus, two different pathways for emission decay. In the first, emission originates with electron trapping by such deep energy traps as anion vacancies {e- + Va f F+ + hν1} to yield electron F-type color centers, whereas in the second, emission originates from electron/trapped hole recombination {e- + OS•- f OS2- + hν2}. The first common step of the pathways is homolytic dissociative chemisorption of hydrogen and methane upon interaction with surface-active hole centers OS•-, produced upon preirradiation of alumina, to give atomic hydrogen H• and methyl radicals CH3•. Thermoprogrammed desorption spectra of photoadsorbed or postsorbed oxygen show that adsorbed oxygen interacts with atomic hydrogen and methyl radicals. The products of thermodesorption were H2O for hydrogen and H2O, CO2, and CH3CH3 for methane. The Solonitsyn memory effect coefficient was also evaluated for oxygen photoadsorption.

Introduction

mental confirmation. Moreover, the emission intimately

Photoactivation of metal oxide surfaces leads to numerous consequences, one of which involves chemical interactions between the surface and the molecules of the gas or liquid phase in a heterogeneous system. Chemisorption of hydrogen-containing molecules, for example, H2, H2O, RH, and ROH, on preirradiated metal oxide surfaces causes the emission of a light pulse following interaction(s) of these molecules with photoinduced surface-active centers. The effect, which we have described as a photoinduced chesorluminescence (PhICL),1,2 was first detected by Andreev and Kotel’nikov.3 In their original study, they examined the fate of H2O, H2, and CH4 on pre-photoactivated surfaces of Al2O3 and MgO3,4 and attributed this PhICL effect to the excited states of surface hydroxyl groups, (OHS-)*, produced by homolytic dissociation of the molecules on photogenerated surface hole centers, OS•-, followed by radiative decay of (OHS-)* to the ground state (eq 1). However, this suggestion later found no experi-

OS•- + RH f (OHS-)* + R•

(1a)

(OHS-)* f OHS- + hν

(1b)

* Corresponding author. E-mail: [email protected]. Fax: (+1) 514-848-2868. † University of St. Petersburg. ‡ Concordia University. § Universita di Pavia. (1) Andreev, N. S.; Emeline, A. V.; Khudnev, V. A.; Polikhova, S. A.; Ryabchuk, V. K.; Serpone, N. Chem. Phys. Lett. 2000, 325, 288. (2) Emeline, A. V.; Polikhova, S. A.; Andreev, N. S.; Ryabchuk, V. K.; Serpone, N. J. Phys. Chem. B 2002, 106, 5956. (3) Andreev, N. S.; Kotel’nikov, V. A. Kinet. Katal. 1974, 15, 1612. (4) Basov, L. L.; Kotel’nikov, V. A.; Solonitsyn, Yu. P. In Spektroskopiya Fotoprevrashenii v Moleculach; Krasnovsky, A. A., Ed.; Nauka: Leningrad, 1977; pp 228-238.

implicated the solid metal oxide specimen in the PhICL emission spectra, which alluded to either energy transfer or charge transfer occurring from the surface to the solid bulk. In a previous study,5 we also observed a PhICL emission caused by hydrogen chemisorption on a prephotoactivated surface of spinel. The emission spectrum was reminiscent of the thermostimulated luminescence spectrum of spinel. However, no emission that could be attributed to the decay of excited (OHS-)* was detected. The PhICL effect was also noticed on pre-photoactivated surfaces of other wide band gap metal oxide surfaces, for example, ZrO2, Ga2O3, Sc2O3, MgO, BeO, and CeO2.6-8 Recently, we reported results of detailed studies1,2 of the process on hydrogen adsorption on preirradiated zirconia that also yields a PhICL emission. It was demonstrated both spectrally and kinetically that this PhICL emission is closely connected to the photo- and (5) Emeline, A. V.; Ryabchuk, V. K. Russ. J. Phys. Chem. 1998, 72, 432. (6) Kotova, O. B. Ph.D. Thesis. Leningrad State University, Leningrad, Soviet Union, 1986. (7) Burukina, G. V. Ph.D. Thesis. Leningrad State University, Leningrad, Soviet Union, 1990. (8) Andreev, N. S.; Kotel’nikov, V. A.; Solonitsyn, Yu. P.; Basov, L. L.; Kuzmin, G. N. Department of Photonics, Institute of Physics, St.Petersburg State University, St.-Petersburg, Russia. Unpublished data.

10.1021/la030228a CCC: $27.50 © 2004 American Chemical Society Published on Web 12/06/2003

130

Langmuir, Vol. 20, No. 1, 2004

thermoluminescence of the solid. The experimental results led to a mechanism to describe the chemical and physical processes that take place in heterogeneous systems (see the following) and that are responsible for the PhICL phenomenon. In the present study, we continue to examine the PhICL effect and propose an analogous model for another heterogeneous system, namely, γ-Al2O3, as the (photo)adsorbent. The photoactivation of the alumina surface causes photochemical interfacial events in the gas-solid heterogeneous system.9-11 For example, the photostimulated adsorption12 of gas molecules such as molecular oxygen, hydrogen, and methane occurs on the solid’s surface. Note that only oxygen is evolved in its original molecular form during the photo- and thermodesorption of reaction products from the surface after the photostimulated adsorption of the corresponding gases. The photoadsorption of hydrogen yields no products in the gas phase during the photoreaction; subsequent heating yields only the desorption of water. The photochemical interaction(s) of methane molecules with the surface of preirradiated alumina causes formation of ethane in the gas phase during the photoreaction, and subsequent thermodesorption yields water, CO2,13 and ethane. Accordingly, alumina behaves as an active photocatalyst.14 In addition to its activity in photostimulated surface reactions, alumina partly retains its photoinduced surface chemical activity after preirradiation in vacuo, owing to the formation of long-lived surface-active centers. Also, it is relevant to mention that alumina was the (photo)adsorbent for which the PhICL effect was observed for the very first time.3 Experimental Section The powdered granulated form of the γ-Al2O3 specimen was of high purity grade having a specific surface area of about 100 m2 g-1 (BET method; nitrogen gas). The specimen was pretreated in a manner identical to that described earlier1,2,5 to remove ubiquitous organic impurities and any other adventitious adsorbed molecules. The reproduction of the original state of the specimens between experiments was achieved by heating the samples in an oxygen atmosphere at 900 K for about 1 h. The experimental errors in the spectral measurements caused by the nonreproducibility of the original state of the specimen did not exceed about (10%. The powdered samples were contained in a quartz cell (path length 3 mm; illuminated area 6 cm2) connected to a high-vacuum setup equipped with an oil-free pump system. The residual gas pressure in the reaction cell was about 10-7 Pa. A Pirani-type manometer (sensitivity, 18 mV Pa-1 for O2 and 24 mV Pa-1 for H2) measured the gas pressures. The monitoring of the chemical composition of the gas phase, partial pressure of the gases, and desorption products were carried out with a mass spectrometer (model MX7301) attached to the reactor. Preirradiation of the solid specimens was achieved with a 120-W high-pressure mercury lamp (MELZ, DRK-120; the photon (9) Yurkin, V. M.; Prudnikov, I. M.; Solonitsyn, Yu. P. Deposited Doc. VINITI 4816-82, 1982; p 107. (10) Kotel’nikov, V. A.; Prudnikov, I. M. Kinet Katal. 1969, 10, 1112. (11) Klimovskii, A. O.; Lisachenko, A. A. Khim. Fiz. 1987, 6, 466. (12) Terms such as (i) photoadsorbed, (ii) photosorbed, and (iii) adsorbed have been used in this paper. The first refers to photostimulated adsorption of a gas molecule during irradiation of the alumina specimen. The second term refers to adsorption of the gas molecule after irradiation was terminated, whereas the term “adsorbed” is the generic word indicating adsorption of the gas molecules onto the alumina specimen. (13) During the thermodesorption stage involving methane, in the oxidative phase there is no doubt that some methanol and other oxidized species also formed in addition to carbon dioxide. Under our experimental conditions, we only observed CO2 as the oxidized product as the intermediates reacted rapidly to mineralization. (14) The alumina acts not only as a photoadsorbent but also as a “photocatalyst” to yield oxidative and reductive products with itself regenerated entirely under the conditions used.

Andreev et al.

Figure 1. Time evolution of hydrogen pressure in the reactor at a constant flow of hydrogen gas (Θ ) 0.140 ( 0.001Pa s-1) into the closed reactor over nonpreirradiated (1) and preirradiated (2) γ-alumina. flux at wavelengths shorter than 250 nm was about 1015 photons cm-2 s-1). The luminescence was recorded with a KSVU-12 spectral setup (LOMO) adapted to record the luminescence emission of solid powders in vacuo. A thermoprogrammed homebuilt heating device was used to carry out the experiments to measure the thermodesorption of preadsorbed gases on the preirradiated surface of Al2O3. The working temperature range of the device was 300-800 K with the rate of linear temperature increase set at 0.3 K s-1.

Results Post-Adsorption of Gas Molecules on Pre-Photoactivated Alumina. As is typical of most metal oxide photocatalysts,9 alumina partly conserves its chemical surface activity induced by photoactivation after irradiation (in vacuo) is terminated. This effect is described as the Solonitsyn memory effect,15 which refers to the postirradiation adsorption event after the termination of irradiation in vacuo. It is characterized quantitatively by the post-adsorption memory coefficient K(t) given by eq 2,16

K(t) )

Npost(t) Nphoto(t)

(2)

where Npost(t) is the number of molecules adsorbed after the termination of irradiation for a given time, t, and Nphoto(t) is the number of photoadsorbed molecules during the same time of irradiation, t. Note that K(t) is a function of the time of preirradiation. The residual activity in the alumina specimen originates from the photoinduced formation of long-lived surface-active centers (similar to the surface color centers) that retain their activity for a time period longer than ∼103 s. Consequently, the addition of gas molecules into the closed reactor containing the preirradiated photocatalyst leads to surface chemical reactions, such as photoinduced chemisorption (postadsorption) of gas molecules (e.g., see Figure 1) and other more complex chemical transformations. Only oxygen is evolved in its original molecular form during thermodesorption of the reaction products from the surface after adsorption of the corresponding gases (see Figure 2). The ratio between the amount of oxygen desorbed after photostimulated adsorption of the oxygen and after post(15) (a) Solonytsin, Yu. P. Russ. J. Phys. Chem. 1958, 32, 1241. (b) Wolkenstein, Th. Electronic Processes on the Semiconductor Surfaces during Chemisorption; Consultants Bureau: New York, 1991. (16) Serpone, N.; Emeline, A. V. Int. J. Photoenergy 2002, 4, 91.

Photoinduced Chesorluminescence Effect

Langmuir, Vol. 20, No. 1, 2004 131

Figure 2. TPD spectra of molecular oxygen on alumina after (1) oxygen post-adsorption on the alumina surface preirradiated in vacuo for 5 min; (2) same as 1 followed by hydrogen postadsorption; and (3) same as 2 followed by oxygen adsorption.

Figure 4. TPD spectra of photoadsorbed molecular oxygen after (1) oxygen photoadsorption on the alumina surface, and (2) same as 1 after post-adsorption of methane.

Figure 3. TPD spectra of photoadsorbed molecular oxygen after (1) oxygen photoadsorption on the alumina surface; (2) same as 1 after post-adsorption of hydrogen; (3) same as 2 after additional adsorption of oxygen; and (4) same as 2 with oxygen present in the gas phase.

Figure 5. TPD spectra of molecular oxygen adsorbed on the nonpreirradiated alumina surface before (1) and after (2) exposure to a hydrogen atmosphere (∼150 Pa) for 30 min.

adsorption of the oxygen when preirradiation in vacuo was carried out for the same time period (5 min) as that for photoadsorption (see Figure 3) gives an estimate of the memory coefficient; K(5 min) ) 0.15. The physical meaning of the memory coefficient is that K(t) reflects that portion of long-lived surface-active centers with respect to the total number of the surface-active centers that formed during the photoactivation of the surface. Therefore, interaction of oxygen with the photoactivated alumina surface can be considered as a simple case of photostimulated or photoinduced chemisorption. However, the interaction(s) of hydrogen and methane molecules with photoactivated surfaces is a much more complex process. The thermodesorption of post-adsorbed hydrogen or methane from the surface causes (at least partly) the formation of oxidation products (viz., water for hydrogen and water and carbon dioxide for methane). Methane adsorption on the pre-photoactivated surface also yields ethane as a product of thermodesorption. Complex mechanisms of chemical interactions of hydrogen and methane on the photoactivated surface were also confirmed by changes observed in the thermoprogrammed desorption (TPD) spectra of preadsorbed oxygen (see, e.g., Figures 2-4). For both photoadsorption and post-sorption of oxygen, additional adsorption of hydrogen and methane caused the disappearance of the adsorbed oxygen, as

evidenced in the low temperature regions (T < 200 C) of the TPD spectra. However, the direct interaction of these gas molecules with adsorbed oxygen is unlikely to play the major role in chemical interactions because the similar regions of oxygen adsorbed on a nonirradiated surface reacts with hydrogen very slowly (see, e.g., Figure 5). Consequently, we infer that oxygen interacts mostly with other, more active species than molecular hydrogen or methane, species that were formed during the secondary interactive steps between molecular hydrogen (or methane) and the photoactivated alumina surface. Atomic hydrogen and methyl radicals produced during the dissociative adsorption of the corresponding molecules on hole surface-active centers are good examples of such reactive species (see Discussion). At the same time, hydrogen adsorption on the preirradiated alumina surface leads to formation of an additional number of electron surface-active centers for oxygen adsorption. The latter is observed as an increase in the amount of adsorbed oxygen on the surface during hydrogen adsorption when oxygen molecules are also in the gas phase or as additional adsorption of oxygen after adsorption of hydrogen on the pre-photoactivated alumina surface (see Figures 2 and 3). The latter case demonstrates that such centers are long-lived adsorption centers, analogous to those responsible for the Solonitsyn memory effect. PhICL Effect. Chemisorption of hydrogen on preformed photoinduced surface hole centers is accompanied

132

Langmuir, Vol. 20, No. 1, 2004

Figure 6. PhICL emission detected at λ ) 540 nm during the adsorption of gases on alumina preirradiated for 15 min at high gas pressures. (1) PhICL emission induced by hydrogen adsorption on alumina preirradiated in vacuo; (2) PhICL emission induced by hydrogen adsorption on alumina preirradiated in oxygen; and (3) PhICL emission induced by methane adsorption on alumina preirradiated in oxygen.

Andreev et al.

Figure 8. Spectrum of the PhICL emission induced by hydrogen adsorption on preirradiated alumina.

Figure 9. Time evolution of the PhICL emission detected at λ ) 340 nm (1; dashed line) and at λ ) 440 nm (2; solid line). Figure 7. Time evolution of the rate of hydrogen adsorption (1; dashed line) and PhICL emission (2; solid line) on alumina preirradiated in vacuo for 15 min at a constant flow of hydrogen (Θ ) 0.140 ( 0.001 Pa s-1) into the closed reactor.

by emission of a light pulse (see Figure 6), which we again attribute to a PhICL luminescence.1 At high gas pressures, the emission decay kinetics are described by a singleexponential decay function; decay time τ ) 0.4 ( 0.1 s at 400 nm. To increase the accuracy of the PhICL emission measurements and to compare the kinetics of the PhICL emission and hydrogen adsorption on preirradiated γ-Al2O3, we measured the kinetics of the processes relative to the hydrogen flow into the reactor introduced at a constant flow rate () 0.140 ( 0.001 Pa s-1). The time evolution of the PhICL emission and the hydrogen adsorption rate are illustrated in Figure 7. The spectrum of the PhICL emission was obtained from the integrated emission at a given wavelength corrected by the sensitivity factor of the photomultiplier tube. The PhICL spectrum presented in Figure 8 corresponds qualitatively to the emission spectrum of photoluminescence of γ-Al2O3 reported earlier.17 Accordingly, we infer that the PhICL emission in γ-Al2O3 is due to energy (and/or charge) transfer from the surface to the solid followed by energy conversion into an emission on the same type of emission centers that are involved in the photoluminescence of the adsorbent alumina. Clearly, the PhICL spectrum is complex, formed by an overlap of at least four single (17) Basov, L. L.; Kotel’nikov, V. A.; Solonitsyn, Yu. P. In Spektroskopia fotoprevrashenii v molekulah; Nauka: Leningrad, 1977; pp 228-238.

emission bands. The complexity of the spectral variation of this PhICL emission is also confirmed by the different kinetic behavior of the PhICL emission at short and long wavelengths (see Figure 9). The emission decay kinetics at the long wavelengths (380-700 nm) follow the time evolution of the hydrogen adsorption rate (Figure 7), whereas at the short wavelengths (330-370 nm), the PhICL decay is faster and emission is complete before the end of hydrogen adsorption. We can conclude, therefore, that the primary step of hydrogen adsorption on preirradiated γ-Al2O3 is the limiting step of the PhICL process and all secondary chemical and physical steps are relatively faster. At liquid nitrogen temperatures and high gas pressures, the PhICL emission intensity is lower (∼60%) than that at ambient temperatures and follows single exponential decay with a shorter decay time, τ ) 0.2 s (see Figure 10). However, the emission can be restored partially by heating the sample in vacuo to ambient temperatures followed by the introduction of molecular hydrogen. This leads to an eightfold increase of the decay time, τ ) 1.6 s. Evidently, there exists some energy heterogeneity of the photoinduced centers for hydrogen adsorption on the surface of γ-Al2O3. Some of these centers are unable to overcome the energy barrier to interact with hydrogen molecules at low temperatures. The PhICL emission is strongly affected by the presence of oxygen either on the surface or in the gas phase. In particular, prior photostimulated adsorption of molecular oxygen on the surface increases the PhICL emission (see Figure 6). The latter correlates with an increase (typical for metal oxides) in the surface concentration of photoinduced active hole centers, namely OS•-, for hydrogen

Photoinduced Chesorluminescence Effect

Langmuir, Vol. 20, No. 1, 2004 133

the pre-photoexcitation of metal oxide semiconductor/ dielectric specimens. Specifically, it can be described as the triggering of physical relaxation processes of the heterogeneous system back to a lower energy state as a result of surface chemical reactions.2 We now examine the general validity of the model proposed previously1,2 that described the processes responsible for the PhICL effect. According to the model, the initial step, which leads to the PhICL emission, is the homolytic dissociative chemisorption of molecular hydrogen (eq 3a) or methane (eq 3b) on the surface-active hole centers (OS•-) yielding highly reactive species such as, for example, atomic hydrogen and methyl radicals, respectively. Figure 10. PhICL emission detected at λ ) 540 nm induced by hydrogen adsorption on preirradiated alumina at high gas pressures: (1) at room temperature, (2) at liquid nitrogen temperatures, and (3) at room temperature after performance of the experiment in 2.

Figure 11. Long wavelength PhICL emission (λ ) 440 nm) induced by hydrogen adsorption at a constant flow of hydrogen (Θ ) 0.140 ( 0.001 Pa s-1) into the reactor without (1) and with (2) oxygen in the gas phase.

chemisorption.18,19 At the same time, the decay time of the PhICL emission (high gas pressure) is decreased to τ ) 0.2 s. (Note that methane adsorption on alumina preirradiated in vacuo does not induce a detectable PhICL emission unless preirradiation is carried out in an oxygen atmosphere.) The presence of oxygen in the gas phase, however, strongly quenches the emission intensity of the PhICL luminescence (see Figure 11). Moreover, gaseous oxygen affects the shorter wavelength emission to a greater extent (i.e., decreases it 10-fold) than the longer wavelength emission (decreases it 2-fold), and also changes the emission decay kinetics. Both require longer times to decay fully. This confirms the implication just described that there exists two different pathways for the induction of the PhICL phenomenon. That is, the pathway for the short wavelength emission differs significantly from the longer wavelength emission. Discussion The PhICL effect represents a fascinating example of the interconnection between chemical and physical relaxation processes in heterogeneous systems caused by (18) Ryabchuk, V. K.; Burukina, G. V. Sov. J. Phys. Chem. 1991, 65, 1621. (19) Emeline, A. V.; Ryabchuk, V. K.; Salinaro, A.; Serpone, N. Int. J. Photoenergy 2001, 3, 1.

OS-• + H2 f OHS- + H•

(3a)

OS-• + CH4 f OHS- + CH3•

(3b)

To the extent that dissociative chemisorption of hydrogen (step 3a) is an activated process, it will have a lower probability of occurring at liquid nitrogen temperatures and will thus increase the survival rate of some of the surface-active centers. Consequently, partial restoration of the PhICL emission at ambient temperatures takes place on subsequent addition of molecular hydrogen (see Figure 10). However, it should be noted that, unlike zirconia, most surface-active hole centers retain their activity even at liquid nitrogen temperatures, inferring that the activation energy barrier for dissociative adsorption of hydrogen on these centers is rather small. Concomitantly, not all active centers are able to cleave homolytically the hydrogen molecule at low temperatures, thus pointing to a nonuniformity of the energy distribution of surface-active centers. As such, the most likely mechanistic scenario involves secondary chemical steps that necessitate highly energetic hydrogen atoms. One such step is interaction of H• species with low-coordinated surface oxygen anions O(LC)S2-.20 The latter entities are a direct source of free electrons (eq 4). Obviously, the radicals formed in steps 3a and 3b can easily react with preadsorbed oxygen (eq 5a,b), causing the changes observed in the TPD spectra of photoadsorbed and post-sorbed oxygen presented in Figures 2-4.

O(LC)S2- + H• f OHS- + e-

(4a)

O(LC)S2- + CH3• f OCH3 S- + e-

(4b)

O2 S•- + H• f {O2H}S- f f products

(5a)

O2 S•- + CH3• f {O2CH3}S- f f products (5b) Other processes that can generate electrons involve the surface radical reactions 6 and 7. Both processes are

OS•- + H• f OHS- + Q

(6)

H•+ H• f H2 + Q′

(7)

accompanied by a release of the corresponding binding energies Q and Q′, which are comparable to the energy of the photons used for pre-excitation of the metal oxide specimen and which can, therefore, also be used to (20) Che, M.; Tench, A. J. Adv. Catal. 1982, 31, 77.

134

Langmuir, Vol. 20, No. 1, 2004

Andreev et al.

generate additional free electrons on alumina (eq 8).

Al2O3 + Q/Q′ f (Al2O3)* f e- + Al2O3

(8)

Accordingly, surface chemical processes occurring on the preirradiated alumina specimen are the primary source of free electrons that initiate the physical relaxation pathway in heterogeneous systems. Subsequent physical steps of such relaxation can be the trapping of free electrons by deep traps, such as anion vacancies (eq 9) or hole centers (by analogy with zirconia)1,2,21 (surfacetrapped holes, OS•-; eq 10) to yield a different PhICL luminescence.

e-+ Va f F+ + hν1

(9)

e-+ OS•- f OS2- + hν2

(10)

Electron trapping can be accompanied by the emission of photons of corresponding energies (hν1 and hν2) that are manifested as different PhICL emissions. Steps 8 and 9 also take place under photoexcitation of the solid to generate free electrons that lead to photoluminescence of the solid. Correspondence of the PhICL emission spectrum with the spectrum of photoluminescence of γ-Al2O3 17 infers that steps 9 and 10 do occur in alumina during chemisorption of hydrogen and methane on the photoinduced surface-active centers. Clearly, the primary sources of free electrons in the PhICL process are the secondary chemical steps involving radical species (steps 4, 6, and 7). On the basis of the just described model, we now discuss the relevant experimental results obtained in this study that add further credence to the postulated model.2 From the data presented in Figure 7, it is obvious that the limiting step of the events resulting in the PhICL phenomenon is hydrogen adsorption on the pre-photoactivated surface of alumina. This seems reasonable because all other steps of the proposed model involve either a highly energetic radical chemistry on the surface or electronic processes of trapping and recombination in the solid, both of which are relatively fast. Note, however, that the short wavelength emission decays in a shorter time, which points to the existence of at least two different emission centers and to additional steps that accelerate the emission decay. To further the discussion of the possible mechanism of the PhICL effect, we consider the results of numerical calculations of the set of differential equations that describe the kinetic behavior of the corresponding steps 3-10 of the proposed model. In this set, the rate of hydrogen adsorption is given by eq 11, whereas in the cae of hydrogen flow into the reactor at the constant flow rate Θ, the pressure in the reactor changes according to eq 12, which determines the pressure in eq 11.

dP(H2ads)/dt ) k3aPH2[OS•-]

(11)

dP/dt ) Θ - k3aPH2[OS•-]

(12)

Then, the rate of change of the concentration of atomic hydrogen formed during the dissociative adsorption of molecular hydrogen on photoinduced active centers, and (21) Emeline, A. V.; Rudakova, A. V.; Ryabchuk, V. K.; Serpone, N. J. Phys. Chem. B 1998, 102, 10906.

Figure 12. Model simulation of the time evolution of the rate of hydrogen adsorption at a constant flow of hydrogen into the reactor (1) and of the PhICL emission through the pathway 1 (2) and pathway 2 (3).

possibly reacting with other surface species, is given by eq 13.

d[H]/dt ) k3aPH2[OS•-] - k6[H•][Os•-] k4a[H•][Os2-] - k7[H•]2 (13) Experimental results demonstrate the existence of at least two different emission sources. By analogy with other heterogeneous systems, we consider as a most plausible mechanism of PhICL emission the trapping of free electrons by deep energy traps such as the anion vacancies. The emission intensity is then described by

Ihν1 ) k9[e-][Va]

(14)

Assuming that the interaction of atomic hydrogen with surface lattice oxygen anions (step 4) is a major source of electrons, the rate of which is given by eq 15, it becomes rather obvious that provided hydrogen adsorption is the slowest step of the mechanism, and the time evolution of the emission through pathway 1 (step 9) will then follow the rate of hydrogen adsorption, in accord with the experimental data (see Figures 7 and 12) because the electronic processes of trapping are very fast and the concentration of the surface lattice oxygen is much greater than the concentration of atomic hydrogen formed through hydrogen adsorption. At the same time, the PhICL

d[e-]/dt ) k4a[H•][Os2-] - k9[e-][Va]

(15)

emission at the shorter wavelengths, occurring through the second emission pathway (step 10), decays faster than the longer wavelength emission and, therefore, faster than the rate of hydrogen adsorption. This infers that there exists a process that competes with the step that causes the emission at the shorter wavelengths. Such competing steps can be electron trapping by surface-active hole centers and interaction of these centers with molecular and atomic hydrogen. Consequently, as the hydrogen pressure increases, the rate of decay of the active centers (eq 16) through the chemical pathway becomes more efficient and prevents the physical decay through electron trapping, thereby causing the PhICL emission of the second sort (hν2) to decay faster (see Figures 9 and 12;

Photoinduced Chesorluminescence Effect

Langmuir, Vol. 20, No. 1, 2004 135

d[e-]/dt ) k4a[H•][OS2-] - k9[e-][Va] k10[e-][OS•-] - k19[e-]PO2 (20) similar role is expected to be played by the CH3• radicals, which causes an absence of detectable PhICL emission after preirradiation in vacuo (eqs 21 and 22). However, strong interaction of the CH3• radicals with photoadsorbed oxygen (step 5b; see also Figure 4) causes the survival of some surface electrons that induce the PhICL emission from methane adsorption, under these conditions.

CH3• + e- f f products

(21)

d[e-]/dt ) k4b[CH3•][OS2-] - k9[e-][Va] k10[e-][OS-] - k21[e-][CH3•] (22) Figure 13. Time evolution of the modeled PhICL emission without (1) and with (2) the presence of oxygen in the gas phase.

eq 16). Note that the decay of surface-active centers

d[OS•-]/dt ) -k3aPH2[OS•-] - k6[H•][OS•-] k10[e-][OS•-] (16) Ihν2 ) k10[e-][OS•-]

(17)

throughelectron trapping and their interaction with atomic hydrogen were previously considered2 to be the major reason for the dependence of the PhICL emission on hydrogen pressure. Moreover, it was also concluded that the process of trapping of free charge carriers by the surface-active centers is responsible for the typical dependence of the rate of heterogeneous photoreactions on the concentration of reagent and on the light irradiance.21 In this latter scenario, the time-dependent changes of the electron concentration follows eq 18.

d[e-]/dt ) k4a[H•][OS2-] - k9[e-][Va] - k10[e-][OS•-] (18) As evident from the experimental data, hydrogen adsorption on the pre-photoactivated alumina surface causes the formation of additional electron surface-active centers for oxygen adsorption. Indeed, the PhICL process is due to electron trapping by anion vacancies, leading to formation of F-type electron color centers (step 9). Those centers located at the surface serve as the long-lived electron surface-active centers for oxygen adsorption. Accordingly, the surface oxidative chemical reaction of hydrogen adsorption initiates the physical relaxation of the solids to yield the trapped electrons (F+ centers; step 9), which ultimately leads to the surface reductive chemical process of oxygen adsorption. In turn, the oxygen present in the gas phase reacts with surface electrons, resulting in a decrease of the surface concentration of electrons and to additional oxygen adsorption (eqs 19 and 20), and therefore to a decrease of the emission intensity in both trapping pathways (steps 9 and 10 and Figure 13). A

O2 + e- f O2•-

(19)

It should be emphasized that the amount of oxygen adsorbed during hydrogen adsorption is greater than the quantity of oxygen adsorbed after hydrogen adsorption. This points to the coexistence of long-lived and shortlived electron surface-active centers as occurs during the photoexcitation of alumina. The magnitude of the memory coefficient K(t) estimated from the TPD spectra is 0.35 (t ) 15 min) in the case of preirradiation of alumina in an oxygen atmosphere and 0.75 if alumina was preirradiated in vacuo. At the same time, more surface-active centers OS•- escape decay through the physical pathway, thus allowing their participation in the chemisorption of hydrogen, the only process that dictates the kinetics of the PhICL emission in such a case. As a result, the time of emission decay becomes longer in accord with the experimental results (see Figure 13). Clearly then, the proposed simplified mechanism describes the kinetic behavior of the PhICL phenomenon observed during the postirradiation adsorption of hydrogen on preirradiated alumina. Concluding Remarks Two major conclusions can be reached on the basis of the experimental results and the proposed mechanism. First, surface chemical reactions such as adsorption of hydrogen and methane, together with subsequent secondary chemical steps, initiate the physical pathway of relaxation of heterogeneous systems that leads to the evolution of the PhICL emission. Second, the chemically initiated physical pathway of relaxation competes with the primary and secondary chemical processes, which involve the generation and decay of surface-active centers on the surface of the (photo)catalyst. It is this competition that determines the kinetic behavior of photoinduced processes in heterogeneous systems, particularly at low gas pressures. Acknowledgment. We are grateful to the Natural Sciences and Engineering Research Council of Canada for a discovery grant (NSERC, Grant A5443 to N.S.) and to the North Atlantic Treaty Organization for a Collaborative Linkage Grant (NATO Grant PST. CLG 979700 with Prof. Ryabchuk of St.-Petersburg State University, Russia, and Prof. Otroshchenko of the Bakh Institute of Biochemistry, Russian Academy of Sciences, Moscow) in support of our work. LA030228A