Identification of Binding Sites for Acetaldehyde Adsorption on Titania

ARTICLE pubs.acs.org/Langmuir

Identification of Binding Sites for Acetaldehyde Adsorption on Titania Nanorod Surfaces Using CIMS Daniel Finkelstein-Shapiro,† Avram M. Buchbinder,†,§ Baiju Vijayan,‡,§,|| Kaustava Bhattacharyya,†,§ Eric Weitz,†,§ Franz M. Geiger,*,†,§ and Kimberly A. Gray*,‡,§ †

Department of Chemistry, Northwestern University, Evanston, Illinois 60208, United States Department of Civil and Environmental Engineering, Northwestern University, Evanston, Illinois 60208, United States § Institute for Catalysis and Energy Processes, Northwestern University, Evanston, Illinois 60208, United States ‡

bS Supporting Information ABSTRACT: The interaction of acetaldehyde with TiO2 nanorods has been studied under low pressures (acetaldehyde partial pressure range 10
4
10
8 Torr) using chemical ionization mass spectrometry (CIMS). We quantitatively separate irreversible adsorption, reversible adsorption, and an uptake of acetaldehyde assigned to a thermally activated surface reaction. We find that, at room temperature and 1.2 Torr total pressure, 2.1 ( 0.4 molecules/nm2 adsorb irreversibly, but this value exhibits a sharp decrease as the analyte partial pressure is lowered below 4  10
4 Torr, regardless of exposure time. The number of reversible binding sites at saturation amounts to 0.09 ( 0.02 molecules/nm2 with a free energy of adsorption of 43.8 ( 0.2 kJ/mol. We complement our measurements with FTIR spectroscopy and identify the thermal dark reaction as a combination of an aldol condensation and an oxidative adsorption that converts acetaldehyde to acetate or formate and CO, at a measured combined initial rate of 7 ( 1  10
4 molecules/nm2 s. By characterizing binding to different types of sites under dark conditions in the absence of oxygen and gas phase water, we set the stage to analyze site-specific photoefficiencies involved in the light-assisted mineralization of acetaldehyde to CO2.

I. INTRODUCTION Organic compounds are naturally present in the environment, arising from metabolic processes. In the atmosphere, they can be degraded photochemically, and in general, the residency times are a few hours.1 However, anthropogenic sources (fossil fuel burning, smoking, aging of materials) can substantially increase the concentration of these molecules above their natural levels and result in adverse health effects.1
3 Airborne pollutants in indoor environments are of particular concern, as they present greater exposure risks and are difficult to remove by common air purification technologies. This problem is exacerbated in enclosed areas with poor circulation giving rise to what is collectively referred to as “sick building syndrome” (SBS).4
10 SBS is quite significant and costs the U.S. 10
30 billion dollars in loss of productivity every year.11,12 High targets for the energy efficiency of buildings impose tighter restrictions on air recirculation, exacerbating the problem. The photocatalyzed degradation of organic compounds to CO2 by inexpensive and readily available semiconductors is a cost-effective, low temperature reaction that can be used for indoor air pollution remediation.13
15 The molecular-level study of the relevant gas
solid interactions, under dark conditions, is a prerequisite for understanding, predicting, and controlling the photocatalytic efficiencies of these materials. Such studies have to be carried out at the partial pressures under which the catalyst is expected to perform because partial pressure plays a crucial role in driving adsorption.16,17 The r 2011 American Chemical Society

typical concentrations of airborne organic pollutants are well below 1 ppm.6,8
10,18 At these partial pressures the interaction of molecules with the surface can differ significantly from the higher pressure range (>100 ppm) usually studied, especially when multiple types of binding sites are available.16,17 For example, only a subset of sites are occupied by adsorbates under low reactant partial pressure conditions.16,19,20 This is illustrated in the case of the interaction of the indoor air pollutant acetone with Degussa P25. El-Maazawi et al. reported acetone surface coverages of 10 molecules/nm2 at a high acetone partial pressure,21 whereas Schmidt et al. reported surface coverages of only 0.01 molecules/nm2 and 0.016 molecules/nm2 for chemisorbed and physisorbed sites, respectively, for partial pressures ranging from 10
7 to 10
4 Torr.19,22,23 This underlines the large difference in surface coverage, 3 orders of magnitude, between high and low reactant partial pressure regimes relevant for indoor air pollution processes. In addition, phase transitions accompanied by reorientation of adsorbed molecules at the surface have also been reported at low reactant partial pressures.24
30 A drastic change in orientation is exemplified by n-heptane which goes from a position where it is lying across the surface at low pressures to one where its long-axis is perpendicular to the surface as pressure Received: July 5, 2011 Revised: September 10, 2011 Published: October 31, 2011 14842

dx.doi.org/10.1021/la2025457 | Langmuir 2011, 27, 14842–14848

Langmuir is increased.24 Because interfacial charge transfer can depend significantly on the geometry of the reactants, these orientational differences may lead to distinct efficiencies of reaction.31 Acetaldehyde is a widely used model compound in testing the photoreactivity of TiO2 catalysts. Furthermore, it is a principal component of SBS gases and also an intermediate in the photodegradation of acetone and ethanol.32,33 Thus, in this work we choose acetaldehyde as our target chemical. Several studies of the interaction of acetaldehyde with TiO2 at high pressures exist. For example, binding studies of acetaldehyde at a pressure of 150 Torr on anatase and rutile nanoparticles using a quartz crystal microbalance found a total number of sites equal to 2.2 and 3.4 molecules/nm2, respectively.34 Experiments investigating submonolayer acetaldehyde coverages on TiO2 surfaces have also been carried out using temperature programmed desorption (TPD). While these studies take place at very low temperatures, they measure a coverage dependent binding energy suggesting a range of binding sites or significant adorbate
adsorbate interactions.35 Complicating these experiments, however, are reactions that occur during the temperature ramp used to make the measurement,36 the conversion of acetaldehyde to crotonaldehyde and other aldol condensates for oxidized surfaces and to butene for reduced surfaces.34,36,37 In particular, the aldol condensation is active at room temperature on rutile surfaces but requires higher temperatures on anatase surfaces. The rates of conversion and deactivation for anatase and rutile have also been investigated in detail at high acetaldehyde partial pressures used in industrial applications.37 In this work we extend the study of acetaldehyde binding into the lower partial pressure range of 10
8
10
4 Torr on TiO2 nanorods.38
41 While we perform our studies in the absence of water other than that present as hydroxyl surface groups and in the absence of oxygen in order to understand this reference behavior, we do our studies in a pressure regime that is environmentally relevant. We probe the surface coverage and partial pressure dependences of irreversible and reversible binding, and determine the binding energy of reversibly bound species. We examine the thermally activated reactions catalyzed by this material and conclude by discussing the difference in adsorption pathways between high and low partial pressures and its important implications for photocatalysis.

II. EXPERIMENTAL SECTION A. Chemicals for CIMS and IR Studies. The analyte for the CIMS experiments was used as received from a dilution of acetaldehyde in helium at 1720 PPM (Airgas). Further dilutions were prepared with a gas manifold in 10 and 3 L glass bulbs from liquid acetaldehyde (Sigma Aldrich Chemicals, ACS reagent >99.5%) and helium gas (Grade 5 UHP). MiliQ (18.2 MΩ) water protonated by a polonium 210 source (NRD in-line Nuclecel ionizer P-2031, 20.0 mCi) was the ionization reagent for the mass spectrometer. The analyte used for IR studies was prepared with the same procedure as used for the CIMS dilutions. B. Nanorod Synthesis and Characterization. Titania nanorods were prepared by a modified hydrothermal method reported previously.38 In a typical synthesis, 2 g of anatase titania powder (assay 99%, Sigma Aldrich Chemicals) was stirred with 50 mL of 10 M NaOH solution (assay 97%, BDH Chemicals) in a 125 mL Teflon cup. The Teflon cup was kept in an oven for 48 h at 120 °C, and the resultant precipitate was washed with 1 M HCl (assay 38%, EMD Chemicals) followed by several washings with deionized water to attain a pH between 6 and 7. The titania nanotube powder thus formed was dried in an oven at 120 °C

ARTICLE

Figure 1. (A) TEM image of titania nanorods (taken from ref 38.). (B) QMS signal during a representative adsorption and desorption experiment. (C) Cartoon of the CIMS flow tube experiment for the two relevant injector positions. overnight. The powders were calcined at 600 °C to obtain titania nanorods that are primarily composed of the anatase phase. The structures were characterized by TEM (JEOL-2100), XRD (Rigaku X-ray diffractometer), EPR, and BET analysis with nitrogen gas (Micromeritics Gemini 2010). The nanorods are cylindrical anatase structures 8 ( 2 nm in diameter and 37 ( 14 nm in length (Figure 1A) with a BET surface area of 64 ( 3 m2/g. While no rutile was detected by XRD, EPR shows that conversion to rutile has started to occur at 600 °C and may account for the collapse of titania nanotubes to form nanorods. At 700 °C, 5.5% of the anatase has transformed to rutile. A detailed description of the nanorods used for the study can be found in the work of Vijayan et al.38 The rods are coated in a slurry and dried with a heat gun at 100 °C. No further pretreatment of the surface prior to the adsorption measurements is carried out. We expect a high density of surface hydroxyl groups. C. Chemical Ionization Mass Spectrometry (CIMS). Adsorption/ desorption measurements were performed in a flow tube coupled to a chemical ionization region where analytes are ionized by collisions with H3O+, and directed to a quadrupole mass spectrometer (QMS). The custom built flow reactor, chemical ionization region, and ion directing optics are described in detail elsewhere,19 and modeled after a design by Molina and co-workers.42
44 Briefly, the catalyst is prepared in a slurry (2 mg/mL). On average, 10 mg of material is deposited onto a 140 g pyrex tube (43 cm length  33 mm i.d.), which was weighed before and after the experiment, with and without catalyst. Mass measurements before and after the experiment consistently agree within 0.7 mg, constituting the largest source of uncertainty in the number of adsorbed species in these 14843

dx.doi.org/10.1021/la2025457 |Langmuir 2011, 27, 14842–14848

Langmuir experiments. Unless specified, the error bars in the figures are twice the mass uncertainty. The catalyst is placed in an outer glass flow tube, which is coupled to the mass spectrometer and fitted on the opposite end with a retractable injector through which the analyte is delivered. During the measurement, a constant flow of helium (250 sccm at 1.2 Torr) is established across the catalyst, providing a flowing background to transport the analyte (flow 0
10 sccm). Flows were set by mass flow controllers (MKS; M100B for He, 1479A for the analyte). The pressure above the catalyst is maintained at 1.2 Torr with a mechanical pump, and the mass spectrometer region is pumped down to 4  10
5 Torr by two BOC Edwards (230 L/s He pumping speed) turbomolecular pumps. All measurements were done at room temperature (293 K). The mass spectrometer (QMS Extrel trifilter spectral focusing assembly quadrupole, 1.2 MHz, 100 W, 1
200 a.m.u.) is controlled by a Merlin Automation software package. Carrier gas flows and the mass spectrometer were turned on the night before each run to allow the baseline to stabilize. A representative trace of the QMS signal is shown in Figure 1B. First, a baseline is established when the injector is closest to the detector, with acetaldehyde maintained at constant partial pressure, bypassing the catalyst coating as indicated in cartoon a. At mark I the injector is retracted to the position indicated in cartoon b (catalyst exposed), and acetaldehyde molecules adsorb on the catalyst surface. Adsorption is manifested as a dip in the QMS signal, and exposure time for these experiments is 30 min. Once steady state is reached, the injector is returned to the front position as indicated in cartoon a (mark II), and molecules desorb from the surface. This is registered as a rise in the QMS signal, and the period of time where the catalyst is left to desorb is 30 min. The number of adsorbed molecules is obtained by integrating the dip of the QMS signal. The baseline used to quantify adsorption is the QMS signal when the catalyst is exposed to the analyte but has reached steady state. The number of desorbed molecules is obtained by integrating the peak of the QMS signal. The baseline for desorption is the steady state signal when the catalyst is not exposed to the analyte. The number of irreversibly bound molecules is calculated by taking the difference between the number of adsorbed molecules and the number of desorbed molecules. The difference in the steady state QMS signal when the catalyst is not exposed to the analyte and when it is exposed to the analyte is taken as a rate of consumption of acetaldehyde associated with a surface reaction. This is an upper bound to the surface reaction rate; it does not take into account slow desorption processes but is also insensitive to slow drifts in the baseline. The complete procedure for determining the rate of reaction and choice of signal baseline is described in detail in the Supporting Information. We note that multiple control experiments were performed to rule out artifacts from the experimental apparatus. D. Infrared Spectroscopy. In situ FTIR absorption spectra were recorded with a Bio-Rad Excalibur FTS-3000 infrared spectrometer equipped with an MCT detector. Each spectrum was obtained by averaging 40 scans at a resolution of 2 cm
1. The reactor, which has been described previously,45 consists of a stainless steel cube, with two CaF2 windows that can be pumped to a base pressure of 1  10
7 Torr. A Baratron capacitance manometer was used to monitor the pressure of the analyte. The exposure of the nanorods was performed by introducing a known concentration of acetaldehyde in helium at fixed pressure in the reactor volume and waiting for equilibrium to be reached. For this study, 10
20 mg of TiO2 nanorods was pressed on a photoetched tungsten grid held between two nickel jaws. The sample was kept under vacuum overnight before each experiment, and all experiments were carried out at room temperature.

III. RESULTS AND DISCUSSION Figure 2 shows the initial adsorption and desorption for three acetaldehyde partial pressures. The difference between the number

ARTICLE

Figure 2. QMS traces at 8.0  10
5, 4.0  10
5, and 1.2  10
5 Torr partial pressure of acetaldehyde, total pressure 1.2 Torr helium. Each measurement is made on fresh catalyst. At minute 12 the injector is placed in the back position, and the analyte adsorbs to the surface. Steady state is reached, and at minute 42 the injector is placed in the front allowing the physisorbed species to desorb. The trace shows the characteristic peak due to desorbing acetaldehyde. The traces have been offset by 5  1015 molecules/s for clarity.

of adsorbed molecules (integrated area of the dip in the signal) and the number of desorbed species (integrated peak in the signal) is indicative of irreversible binding. We also note that even after 30 min the baseline has not recovered to its initial value. This behavior is consistent with a surface reaction that consumes the adsorbed acetaldehyde. We conclude that the interaction of acetaldehyde with the titania nanorods is described by three processes: chemisorption characterized by irreversible adsorption, physisorption characterized by reversible adsorption, and a surface reaction. In the text that follows, we quantify these three processes by the CIMS technique and identify the thermal surface reaction products using FTIR. A. Irreversible Binding. At the highest partial pressure investigated, 8.0  10
4 Torr, the surface coverage of irreversibly bound acetaldehyde is 2.1 ( 0.4 molecules/nm2 (Figure 3A), in agreement with the measurements of Rekoske and Barteau performed at 150 Torr.34 A considerably lower site density previously measured for acetone (0.010 molecules/nm2) highlights the diverging behavior that can occur at low (10
4 Torr) versus high (>1 Torr) pressures of closely related molecules.19 Our measurements show that there exists a steep pressure dependence for irreversible binding (Figure 3A): Below 1.2  10
5 Torr, the surface coverage of chemisorbed species does not exceed 0.30 ( 0.06/nm2 (14% of irreversible coverage at higher pressures) while above 2.8  10
5 Torr, full saturation coverage (taken as the constant coverage at the highest range investigated) is achieved. This is quite interesting given that all physisorption sites (reversible binding, see below) are already saturated at lower partial pressures (10
5 Torr). To test whether longer exposure times increase the surface coverage for chemisorbed species at 1.2  10
5 Torr partial pressure of acetaldehyde, we repeated adsorption cycles a total of 3 times, corresponding to a total exposure of 90 min (Figure 3B). We observe that additional irreversible binding sharply decreases after the first two cycles and that the total surface coverage of irreversibly bound species at low partial pressures for three cycles amounts to only 0.47 molecules/nm2 (adsorption measurements performed on 9 cycles show a value of 0.57 molecules/nm2). 14844

dx.doi.org/10.1021/la2025457 |Langmuir 2011, 27, 14842–14848

Langmuir

ARTICLE

Figure 4. Adsorption isotherm for the reversibly bound acetaldehyde molecules (b) and the best Langmuir fit to the data (—) and log
log plot of the data (inset).

Figure 3. (A) Total number of irreversibly bound acetaldehyde molecules on fresh TiO2 nanotubes during a 30 min exposure to acetaldehyde as a function of acetaldehyde partial pressure. (B) Number of irreversibly bound acetaldehyde molecules for 3 consecutive adsorption cycles for a pressure below saturation of the surface (1.2  10
5 Torr, b) and above saturation (8.0  10
5 Torr, O). The value reported is the number of acetaldehyde molecules adsorbed during that cycle.

The situation is similar for an acetaldehyde exposure of 8  10
5 Torr. Because chemisorbed molecules have stronger interactions with the surface, it is likely they also exhibit better electronic coupling for interfacial charge transfer. This marked acetaldehyde partial pressure dependence in the surface coverage of chemisorbed species at this low operating pressure may dictate the efficiency of photo-oxidation. A pressure dependence for irreversible adsorption could be explained by an energy barrier to reach certain sites. In this case, increasing the pressure results in a larger chemical potential of the gas phase species, facilitating the crossing of the barrier. This change in chemical potential is given by46   3 P2 Δμ ¼ kT ln 2 P1 where k is Boltzmann’s constant, T the temperature, and P2 and P1 the final and initial pressures, respectively. For a change in pressure from 1.2  10
5 to 2.8  10
5 Torr, the chemical potential has a value of 1.275kT. This energy is also readily provided by thermal motion, and so it is unlikely that the small change in chemical potential has such a drastic effect on the population of sites. It is possible collective effects such as a reorientation of the molecules can explain our observations. An effective 2D phase transition would reorient the adsorbates to a

position that leads to more irreversible binding. This interesting point will be pursued further in a separate study. It is also possible that some of the adsorbed species measured as irreversibly adsorbed species react at the surface to form aldol condensates. We elaborate more on this point in section III.C. B. Reversible Sites. Sites that mediate acetaldehyde physisorption are most easily investigated once all the sites for irreversible binding have been saturated by pre-exposure of the catalyst to 4.0  10
5 Torr acetaldehyde. Under these conditions, the total number of physisorbed molecules at saturation is 0.09 ( 0.02 molecules/nm2, only 4% of all surface bound acetaldehyde. Using a Langmuir adsorption model for these reversibly bound adsorbates in the 10
8
10
4 Torr region, we calculate a binding energy of 43.8 ( 0.2 kJ/mol referenced to a standard state of 1 atm (Figure 4, uncertainty is the error of the fit). This value is close to the binding energy of acetone on Degussa P25 and is consistent with multiple hydrogen bonding interactions.19 This behavior stands in marked contrast to what is reported under very high partial pressures where additional monolayers and sites continue to be filled as the pressure is increased, with much lower binding energies.47,48 Multilayer formation has not begun at the very low partial pressures investigated in this work, as evidenced by the surface saturation coverage and shape of the isotherm shown in Figure 4. Precovering the surface before measuring the isotherm facilitates the measurement of the reversibly bound molecules although intermolecular interactions are likely to affect the binding strength. An additional study is needed to measure the binding strength of reversible molecules as a function of coverage of irreversibly bound or oligomerized molecules. C. Thermal Reaction. As mentioned previously, the acetaldehyde signal does not recover to the original baseline when the analyte is flowing over the catalyst. This small shift in the steady state QMS signal is indicative of a chemical reaction at the surface. 49 While reaction products could desorb from the surface, we did not detect gas phase species other than acetaldehyde in the 20
150 mass range: either the concentration of gas phase products was below our detection limits (the products could have low ionization efficiencies which for CIMS using H3 O+ reagent ions scales with proton affinity), or all reaction products remained adsorbed to the surface. These species may be higher molecular weight species, and 14845

dx.doi.org/10.1021/la2025457 |Langmuir 2011, 27, 14842–14848

Langmuir

ARTICLE

Table 1. IR Assignments for Acetaldehyde on TiO2 Nanorods for a Low Dosage of 4  10
6 Torr l and a High Dosage of 2.4  10
2 Torr l low dosage (cm‑1) high dosage (cm‑1)

3260 2914

Figure 5. (A) FTIR spectra of TiO2 nanorods exposed to 4  10
6 Torr l acetaldehyde and (B) 2.4  10
2 Torr l. For both doses we show (a) the background taken as fresh TiO2 nanorods, (b) the spectrum during exposure, and (c) the spectrum after evacuation of the chamber with a turbomolecular pump.

could be more difficult to degrade, and, as such, may deactivate the surface during photocatalysis. Figure 5A,B shows the spectra for acetaldehyde on TiO 2 nanorods for two doses of acetaldehyde corresponding approximately to 4  10
6 Torr l and 2.4  10
2 Torr l, respectively. Assuming a sticking coefficient of 1 and 15 mg of material, 4  10
6 Torr l corresponds to 6.1  10
5 L, and 2.4  10
2 Torr l corresponds to 0.37 L. For both doses, spectrum a is the background consisting of fresh TiO 2 nanorods, spectrum b shows the catalyst after dosing the surface, and spectrum c shows the catalyst after dosing and evacuating the chamber with a turbomolecular pump. The low-dosage spectrum shows that only acetaldehyde is present, on the basis of the band assignment in the literature (Table 1).34,50 The high-dosage spectrum shows ro-vibrational bands of gas phase water around 3750 and 1650 cm
1, which disappear upon evacuation. This spectrum also shows a strong increase of the Ti
OH stretching signal around 3400 cm
1 and the appearance of a peak assigned to the CdC double bond stretch of crotonaldehyde50 or higher order aldol condensates at 1627 cm
1. These observations are consistent with the generation of aldol condensates (reaction 1 shows the aldol condensation for acetaldehyde to crotonaldehyde), which is a reaction known to be catalyzed by rutile at room temperature and anatase at higher temperatures.34,37,51

assignment34,50

3368

n.a.

3260

n.a.

Ti—OH Ti—OH

2960

νas(CH3)

acetaldehyde

2918

νs(CH3)

acetaldehyde

2849 2083

2νsA0 Fermi

acetaldehyde carbon monoxide

1627

ν(CdC)

crotonaldehyde

1573

1551

νas(COO)

acetate

1360

1358

δ(CH)

acetaldehyde

1262

δ (C—OH) 3-hydroxybutenal

We conclude that particular defect sites associated with the TiO2 nanorods, or the incipient amounts of interfacial rutile present in the nanorods, contribute to generating a more active surface at room temperature than what is known for anatase. Figure 5Bc shows that the aldol condensation products remain on the surface, which may potentially deactivate the catalyst during photooxidation and slow down acetaldehyde mineralization. Vibrational bands assigned to the COO
of acetate or formate, 3-hydroxybutanal, and surface-bound CO are also observed (Table 1). 3-Hydroxybutanal may be an intermediate in the aldol condensation process, while the COO
and CO species indicate oxidative adsorption and subsequent dissociation of acetaldehyde is taking place. The products of this reaction desorb once the chamber is evacuated. The presence of oxygen, as would occur under operating conditions of these catalysts, would be expected to increase the rate of this reaction. By performing nine adsorption/desorption cycles at an acetaldehyde partial pressure of 1.2  10
5 Torr using CIMS, we investigated the progression over time of the rate of acetaldehyde consumed by the reaction. Figure 6A shows the reaction rate, which was obtained from the difference of the steady state after adsorption to the steady state after the following desorption (see Supporting Information for a detailed description of the calculation of the rate of reaction). The data clearly show that the rate of the reaction decreases with time and reaches steady state around 0.5
1.0 molecules nm
2 s
1. We attribute this finding to the occupation of all sites available for the conversion of acetaldehyde to other molecules, consistent with a portion of the products remaining bound to the surface. The uncertainty includes the uncertainty in the mass measurement and the uncertainty from referencing the signal to the baseline after desorption (see Supporting Information). Figure 6B shows the number of acetaldehyde molecules adsorbed and desorbed from the catalyst for each cycle, excluding the uptake of molecules by the thermal reactions. An analysis of the reaction rate at partial pressures that do not saturate the irreversible binding sites and at partial pressures that do saturate the irreversible sites reveals similar behavior (see Supporting Information, S3). If we quantify the amount of material adsorbed in this type of process, we see that it amounts to more material than the measured saturation coverage for irreversibly bound species. In the case of the nine consecutive exposures at 1.2  10
5 Torr acetaldehyde partial pressure we find that over all cycles an additional 3.5 molecules/nm2 have been consumed by the surface and converted to other products. While we cannot rule out 14846

dx.doi.org/10.1021/la2025457 |Langmuir 2011, 27, 14842–14848

Langmuir

ARTICLE

but the difference in surface coverage exceeds 2 orders of magnitude.19 This points to factors beyond steric adsorbate
adsorbate effects, such as different modes of adsorption for the two species. It is possible orientational effects, discussed in the Introduction, are causing this large difference in surface coverage. Although these studies were conducted on different types of titania surfaces, we expect, on the basis of previous studies, that Degussa P25 (75% anatase, 25% rutile, surface used to study acetone binding) would have a higher binding site density than the titania nanorods: rutile has a higher binding site density than anatase and P25 has a higher proportion of rutile than the nanorods.34 This makes the expected difference between the binding site density of acetaldehyde and acetone on nanorods even more pronounced. Diverging behavior at low pressures for two species that exhibit comparable binding densities at high partial pressures suggests that the chemistry and reaction pathways of more complex organic molecules will change as a function of pressure, depending on the surface affinity of the intermediates. These phenomena will be the subject of future work.

Figure 6. (A) Reaction rate of aldol condensation of acetaldehyde on TiO2 nanorods. Each point represents the difference between steady state after adsorption, where the thermal reaction is mainly responsible for acetaldehyde uptake by the surface, to the steady state after the following desorption, which is the best estimate for the baseline (number reported is the absolute magnitude). (B) Number of adsorbed (gray) and desorbed (white) acetaldehyde molecules for TiO2 nanorods exposed to a pressure of 1.2  10
5 Torr of acetaldehyde for the indicated amount of time.

that a portion of the irreversibly bound species (see III.A) also undergo oligomerization reactions, the difference in rates for these two processes suggests different mechanisms are at their root. D. Consequences for Photocatalysis at Environmentally Relevant Partial Pressures. The Langmuir
Hinshelwood (L-H) equation often describes the photodegradation of organic compounds well.52,53 In the case of acetaldehyde at low partial pressures, deviations from the L-H model are expected given the changes in adsorption properties: the number of accessible chemisorption sites becomes limited below 10
4 Torr acetaldehyde partial pressure (10 ppb when referenced to one atmosphere, a typical concentration found in the environment) and no multilayer formation is observed, in contrast with high partial pressures. Operating below 10
4 Torr (the most environmentally relevant partial pressure range18), the surface coverage of irreversibly bound acetaldehyde depends strongly on the acetaldehyde partial pressure. If this behavior holds in the presence of oxygen and moisture, this would reduce the number of reactants available for charge transfer in a photoreaction. E. Role of the Methyl Group. Given that the dipole moments of acetaldehyde (2.75 D54) and acetone (2.93 D55) are similar, but that acetone has an additional methyl group, we would expect lower acetone surface coverages on TiO2. This is indeed observed,

IV. CONCLUSION Above 10
4 Torr partial pressure, 96% of surface-bound acetaldehyde is chemisorbed on titania nanorods, with 2.1 ( 0.4 molecules/nm2. The amount of chemisorption exhibits a strong partial pressure dependence between 10
5 and 10
4 Torr, indicating an upper limit to the useful operating pollutant partial pressure if used for air quality control. We measure an enthalpy of adsorption for the reversibly bound molecules of 43.8 ( 0.2 kJ/mol consistent with multiple hydrogen bonds. We measure, on a slower time scale, an uptake of acetaldehyde which we have determined to be a combination of aldol condensation and oxidative binding with surface OH groups. The reaction, with an initial rate of 7 ( 1  10
4 molecules/nm2s, slows down with time indicating a deactivation of the sites that thermally catalyze these reactions. The CIMS allows us to monitor the density of reversible binding sites as the thermal reactions proceed, and we find that these remain unaffected. We conclude that the sites for reversible binding are distinct from those that catalyze the surface chemistry of acetaldehyde on nanorods. The contrast offered by high pressure studies, as well as high and low pressure studies for the adsorption of acetone on TiO2, shows that the chemistry can be significantly different at low pressures and in this region is more sensitive to the type of molecule studied. By continuing this work under conditions where the catalyst is irradiated with UV light in the presence of oxygen, we will study the efficiencies of each one of these binding sites for photocatalysis. Such studies are of fundamental interest as they bring us closer to understanding heterogeneous photocatalysis at the level of the active site. ’ ASSOCIATED CONTENT

bS

Supporting Information. Time constant for the initial adsorption at several acetaldehyde partial pressures. Detailed procedure and discussion for determining the reaction rates using CIMS. Reaction rate for low and high pressures. Additional figures. This material is available free of charge via the Internet at http://pubs.acs.org.

’ AUTHOR INFORMATION Corresponding Author

*E-mail: [email protected] (K.A.G.), f-geiger@northwestern. edu (F.M.G.). 14847

dx.doi.org/10.1021/la2025457 |Langmuir 2011, 27, 14842–14848

Langmuir )

Present Address

Center for Materials for Electronics Technology, Kerala, India 680581.

’ ACKNOWLEDGMENT Sample characterizations (XRD, TEM, FTIR) were performed in the JB Cohen X-ray facility and NUANCE at Northwestern University. D.F.-S. and K.A.G. thank Honeywell, Inc., for support. B.V., A.M.B., K.B., E.W., and F.M.G. performed the work under the auspices of the U.S. Department of Energy, Contract DEFG02-03 ER 15457/A003 and DE-AC02-06CH11357 (ICEP). This work is also supported by National Science Foundation Atmospheric Chemistry Division under Grant NSF ATM-0533436 (F.M.G.). F.M.G. gratefully acknowledges an Irving M. Klotz professorship. ’ REFERENCES (1) Finlayson-Pitts, B. J.; Pitts, J. N. Chemistry of the Upper and Lower Atmosphere: Theory, Experiments, and Applications; Academic Press: New York, 2000. (2) Bailis, R.; Ezzati, M.; Kammen, D. M. Science 2005, 308, 98. (3) Jenkins, P. L.; Phillips, T. J.; Mulberg, E. J.; Hui, S. P. Atmos. Environ., Part A 1992, 26, 2141. (4) Housing: Sick Building Syndrome (SBS), no. 2; World Health Organization; Regional Office for Europe; Vol. 2004. (5) Indoor Air Facts No. 4 (revised): Sick Building Syndrome (SBS), revision ed.; United States Environmental Protection Agency; Office of Radiation and Indoor Air (6609J), 1991; Vol. 2004. (6) Mølhave, L. Environ. Int. 1989, 15, 65. (7) In American Society of Heating, Refrigerating and Air-Conditioning Engineers (ASHRAE) Position Papers: Indoor Air Quality; Atlanta, GA, 2001. (8) Morrison, G. Environ. Sci. Technol. 2008, 42, 3494. (9) Shaughnessy, R. J.; McDaniels, T. J.; Weschler, C. J. Environ. Sci. Technol. 2001, 35, 2758. (10) Weschler, C. J.; Shields, H. C. Atmos. Environ. 1997, 31, 3487. (11) Fisk, W. J. Annu. Rev. Energy Environ. 2000, 25, 537. (12) Fisk, W. J.; Rosenfeld, A. H. Indoor Air 1997, 7, 158. (13) Fujishima, A.; Rao, T. N.; Tryk, D. A. J. Photochem. Photobiol., C 2000, 1, 1. (14) Fujishima, A.; Zhang, X. C.R. Chim. 2006, 9, 750. (15) Kormann, C.; Bahnemann, D. W.; Hoffmann, M. R. J. Phys. Chem. 2002, 92, 5196. (16) Masel, R. I. Principles of Adsorption and Reaction on Solid Surfaces; Wiley: New York, 1996. (17) Boer, J. H. d. The Dynamical Character of Adsorption; Oxford University Press: Clarendon, 1968. (18) Baez, A. P.; Belmont, R.; Padilla, H. Environ. Pollut. 1995, 89, 163. (19) Schmidt, C. M.; Weitz, E.; Geiger, F. M. Langmuir 2006, 22, 9642. (20) Somorjai, G. A. Chemistry in Two Dimensions; Cornell University Press: Ithaca, 1981. (21) El-Maazawi, M.; Finken, A. N.; Nair, A. B.; Grassian, V. H. J. Catal. 2000, 191, 138. (22) Schmidt, C. M.; Buchbinder, A. M.; Weitz, E.; Geiger, F. M. J. Phys. Chem. A 2007, 111, 13023. (23) Schmidt, C. M.; Savara, A.; Weitz, E.; Geiger, F. M. J. Phys. Chem. C 2007, 111, 8260. (24) Jura, G.; Loeser, E. H.; Basford, P. R.; Harkins, W. D. J. Chem. Phys. 1945, 13, 535. (25) Jura, G.; Harkins, W. D.; Loeser, E. H. J. Chem. Phys. 1946, 14, 344. (26) Jura, G.; Loeser, E. H.; Basford, P. R.; Harkins, W. D. J. Chem. Phys. 1946, 14, 117. (27) Jura, G.; Criddle, D. J. Chem. Phys. 1951, 55, 163. (28) Shereshefsky, J. L.; Weir, C. E. J. Am. Chem. Soc. 1936, 58, 2022.

ARTICLE

(29) Shereshefsky, J. L.; Weir, C. E. J. Phys. Chem. 1956, 60, 1162. (30) Adamson, A. W. Physical Chemistry of Surfaces, 5th ed.; John Wiley & Sons: New York, 1990. (31) Marcus, R. A. Rev. Mod. Phys. 1993, 65, 599. (32) Masih, D.; Yoshitake, H.; Izumi, Y. Appl. Catal., A 2007, 325, 276. (33) Coronado, J. M.; Kataoka, S.; Tejedor-Tejedor, I.; Anderson, M. A. J. Catal. 2003, 219, 219. (34) Rekoske, J. E.; Barteau, M. A. Langmuir 1999, 15, 2061. (35) Zehr, R. T.; Henderson, M. A. Surf. Sci. 2008, 602, 2238. (36) Luo, S.; Falconer, J. L. J. Catal. 1999, 185, 393. (37) Rekoske, J. E.; Barteau, M. A. Ind. Eng. Chem. Res. 2011, 50, 41. (38) Vijayan, B.; Dimitrijevic, N. M.; Rajh, T.; Gray, K. J. Phys. Chem. C 2010, 114, 12994. (39) Cozzoli, P. D.; Kornowski, A.; Weller, H. J. Am. Chem. Soc. 2003, 125, 14539. (40) Kolen’ko, Y. V.; Kovnir, K. A.; Gavrilov, A. I.; Garshev, A. V.; Frantti, J.; Lebedev, O. I.; Churagulov, B. R.; Van Tendeloo, G.; Yoshimura, M. J. Phys. Chem. B 2006, 110, 4030. (41) Grassian Vicki, H. In Nanoscale Materials in Chemistry: Environmental Applications; American Chemical Society: Washington, DC, 2010; Vol. 1045, pp 15
33. (42) Percival, C. J.; Smith, G. D.; Molina, L. T.; Molina, M. J. J. Phys. Chem. A 1997, 101, 8830. (43) Poschl, U.; Canagaratna, M.; Jayne, J. T.; Molina, L. T.; Worsnop, D. R.; Kolb, C. E.; Molina, M. J. J. Phys. Chem. A 1998, 102, 10082. (44) Bertram, A. K.; Ivanov, A. V.; Hunter, M.; Molina, L. T.; Molina, M. J. J. Phys. Chem. A 2001, 105, 9415. (45) Yeom, Y. H.; Wen, B.; Sachtler, W. M. H.; Weitz, E. J. Phys. Chem. B 2004, 108, 5386. (46) Reif, F. Fundamentals of Statistical and Thermal Physics; McGraw-Hill: New York, 1967. (47) Ohko, Y.; Tryk, D. A.; Hashimoto, K.; Fujishima, A. J. Phys. Chem. B 1998, 102, 2699. (48) Kim, H.; Choi, W. Appl. Catal., B 2007, 69, 127. (49) Jayne, J. T.; Poschl, U.; Chen, Y.-m.; Dai, D.; Molina, L. T.; Worsnop, D. R.; Kolb, C. E.; Molina, M. J. J. Phys. Chem. A 1997, 101, 10000. (50) Singh, M.; Zhou, N.; Paul, D. K.; Klabunde, K. J. J. Catal. 2008, 260, 371. (51) Idriss, H.; Barteau, M. A. Catal. Lett. 1996, 40, 147. (52) Peral, J.; Ollis, D. F. J. Catal. 1992, 136, 554. (53) Alberici, R. M.; Jardim, W. F. Appl. Catal., B 1997, 14, 55. (54) Turner, P. H.; Cox, A. P. J. Chem. Soc., Faraday Trans. 2 1978, 74, 533. (55) Peter, R.; Dreizler, H. Z. Naturforsch., A 1965, 20, 301.

14848

dx.doi.org/10.1021/la2025457 |Langmuir 2011, 27, 14842–14848