Heterogeneous Photocatalysis and Prospects of TiO₂-Based Photocatalytic DeNOxing the Atmospheric Environment

Abstract

This article reviews the efforts of the last two decades to deNOxify the atmospheric environment with TiO₂-based photocatalytic materials supported on various cementitious-like substrates. Prior to undertaking this important aspect of applied photocatalysis with metal-oxide emiconductor photocatalysts, however, it is pertinent to describe and understand the fundamentals of Heterogeneous Photocatalysis. The many attempts done in a laboratory setting to degrade (deNOxify) the major components that make up the NO_x, namely nitric oxide (NO) and nitrogen dioxide (NO₂), but most importantly the efforts expended in deNOxifying the real environment upon depositing titania-based coatings on various model and authentic infrastructures, such as urban roads, highway noise barriers, tunnels, and building external walls among others, are examined. Both laboratory and outdoor experimentations have been performed toward NO_x being oxidized to form nitrates (NO₃⁻) that remain adsorbed on the TiO₂-based photocatalytic surfaces (except in tunnels—indoor walls) but get subsequently dislodged by rain or by periodic washings of the infrastructures. However, no serious considerations have been given to the possible conversion of NO_x via photocatalytic reduction back to N₂ and O₂ gases that would restore the atmospheric environment, as the adsorbed nitrates block the surface-active sites of the photocatalyst and when washed-off ultimately cause unduly damages to the environment.

1. Introduction

The atmosphere is a very complex matrix that, in addition to nitrogen (N₂) and oxygen (O₂) gases, so important to human survival, consists of various other pollutant gases albeit at much lower concentrations: (i) carbonaceous oxides (CO_x); (ii) nitrogen oxides (NO_x); (iii) sulfur oxides (SO_x); (iv) various hydrocarbons (HCs); and (v) particulate matter. These pollutants originate from both natural sources (e.g., volcanic eruptions, wildfires, lightning, and natural degradation of forests among others) and anthropogenic areas (fertilizers and livestock, farms, and urban areas), stationary sources (e.g., industries, power plants, and sewage treatment plants), and mobile sources (e.g., automobiles, trucks, buses, motorcycles, ships, and airplanes) (see Figure 1) [1]. The natural sources of chemical pollutants, however, are of lesser concerns as they are part of the natural environment equilibrium, contrary to the anthropogenic sources that keep increasing in number and concentration with the ever increasing global human population and society’s continuous increasing demand for energy and associated technological advances.

Undoubtedly, the most important sources of air pollutants implicate the combustion of fossil fuels to produce energy (residential heating and electricity-generating power plants), together with major metallurgical industries, cement/construction industries, and the transportation sector. Figure 1 also identifies the primary pollutants from various sources: carbon monoxide, sulfur dioxide, ammonia, volatile organic compounds (VOCs), particulates and, relevant to the present article, the two major NO_x agents (NO and NO₂). Subsequently, through various interacting events that involve the Sun’s radiation, secondary pollutants are generated, among which are sulfur trioxide, ozone, hydrogen peroxide, and sulfuric and nitric acids (the causes of acid rain). Another class of air pollutants generated from internal combustion engines and industrial fumes that react in the atmosphere with sunlight produce secondary pollutants that, in combination with the primary emissions, create photochemical smog [2].

Pre-industrial concentrations of atmospheric nitrogen oxides have increased steadily from about 280 ppbv to ca. 320 ppbv until a decade ago (2010), with estimated annual emissions of 13.8 Tg of N per year (teragrams; 10¹² g), of which ca. 70% is produced by nitrification and denitrification processes in undisturbed terrestrial environments and world’s oceans, and ca. 3 Tg of N per year (~8%) from agricultural tillage, fertilizer use, and animal wastes [3].

The NO_x gases are formed in large measure in gasoline/diesel combustion engines and in power plants that use fossil fuels to produce electricity via high-temperature combustion/oxidation of the fuel’s nitrogen with air oxygen. Initially, only NO is formed followed by formation of NO₂ after combustion in the exhaust and in the atmosphere in the presence of more O₂. Figure 2a reports the 2011 levels of NO_x emissions in the European Union [4], while Figure 2b reports the 2005 NO_x emission levels in the United States [5]. The major anthropogenic sources of nitrogen oxides are combustion engines (transportation sector) and the electricity/heating sectors.

Most of the tropospheric ozone is formed when NO_x, CO and VOCs react in the atmosphere in the presence of sunlight, and, although they might originate in urban areas, airstreams can carry the NO_x far from its sources causing ozone formation in less populated regions. Globally, a VOC whose atmospheric concentration has increased greatly during the last century (viz., methane) contributes to the formation of ozone [6]. A series of complex reactions that involve a VOC (e.g., CO) in the formation of ozone implicates oxidation of this VOC by a hydroxyl radical (^•OH) [7] first to yield the radical species HO^•CO (Equation (1)), which subsequently reacts with oxygen to produce the hydroperoxy radical HOO^• (Equation (2)) that later reacts with NO to give NO₂ (Equation (3)); the latter photolyzes in sunlight to NO and atomic O(³P) (Equation (4)), which by reaction with oxygen yields ozone (O₃; Equation (5)).

OH + CO → HO^•CO

(1)

HO^•CO + O₂ → HOO^• + CO₂

(2)

HOO^• + NO → ^•OH + NO₂

(3)

NO₂ + hν → NO + O(³P)

(4)

O(³P) + O₂ → O₃

(5)

While the chemistry involving other VOCs might be more complex, the critical step that leads to ozone formation remains nonetheless the oxidation of NO to NO₂ by HOO^• radicals. Nitrogen dioxide also reacts with hydrocarbon molecules present in VOCs to produce yet another pollutant (peroxyacetyl nitrates; PAN), a component of photochemical smog that is mostly responsible for eye irritation and is more damaging to plants than ozone [8].

The NO_x family of pollutants (NO, NO₂, N₂O, and their derivatives) causes a wide range of health issues. Nitric oxide (NO) spreads to all parts of the respiratory system because of its low solubility in water, while the health effects of NO₂ are related to its ability to dissolve in moisture to produce HNO₃ acid—a strong mineral acid. Some of the acute health effects include eye irritation (stinging and watering), throat irritation (pungent smell, stinging nose, and coughing), lung irritation (coughing, wheezing and tight chest—difficulty in breathing), and asthma triggered in asthmatics [9,10]. The most serious acute effects occur after significant exposure to NO₂ causing: (a) acute pulmonary edema—fluid from damaged lung tissue pours out into air spaces preventing air from getting to deeper lung thereby causing choking (asphyxia); and (b) other chronic health effects such as asthma and obliterative bronchiolitis, in which the smallest air passages (the bronchioles) are seriously scarred and become distorted and blocked. Consequently, no one questions the need for NO_x-free clean air as essential to maintain/enhance an individual’s health, and to maintain the integrity of the surrounding environment. In this regard, transformation of the two major NO_x species (NO and NO₂), indeed their suppression, has become a necessity as they underwrite (with the VOCs) the formation of hazardous secondary air pollutants and the accompanying photochemical smog.

Two cities where photochemical smog is not insignificant are the Greater Los Angeles (LA) area in the United States and Beijing, China. Home to nearly 19 million people and located in a geological basin confined by the Pacific Ocean and mountains, LA is the basin of considerable pollution caused by its car-centric culture, its bustling industries and ports, its sprawling development, and its sunny climate with often stagnant winds. It was only in the 1950s that hydrocarbons and NO_x were recognized as the source of photochemical smog (Figure 3); however, with the implementation of mandatory catalytic converters in automobiles in the last two decades, smog has been attenuated somewhat.

With its rapid growth and home to nearly 20 million people, and being an important industrial hub, Beijing is a city where poor air quality has been for decades a regrettable fact of everyday life owing to the presence of significant quantities of particulate matter and photochemical smog, as (Figure 4) experienced by the author in the early 1990s; in subsequent trips several years later, however, this author experienced significant improvements of air quality but by no means have the pollutants and smog been totally eliminated.

There have been many attempts to remediate the occurrence of NO_x and VOC species in polluted urban environments with TiO₂-based photocatalytic cementitious-like materials and photocatalytic coatings (paints) on various supports [11]. Several studies report on the performance of titania deposited on, or otherwise incorporated into cementitious substrates toward the minimization, if not suppression, of air pollutants (see, for example, Refs. [12,13,14,15,16,17,18]). Laboratory studies have shown, rather conclusively, that NO_x can be oxidized to nitrate anions [19,20], while VOCs can be converted into CO₂ and H₂O [21]. Of some concern, however, are studies that demonstrate the formation of harmful intermediates (e.g., nitrous acid, HONO), which are far more harmful to human health than either NO or NO₂ during the disposal of NO_x [22,23]. Not least is the potential that nitrates (NO₃⁻) produced and deposited on the TiO₂ particulate surface in the disposal of NO_x may be implicated in reNOxification reactions; that is, back to NO_x [24,25,26] and formation of ozone [26] that would forestall the application of TiO₂-based photocatalytic surfaces to improve the quality of urban air environments.

The objective of this review article is to examine the various attempts at eliminating NO_x species in the urban environment produced mostly by vehicular traffic through application of commercially available titania-based photocatalytic materials, coatings and paints in tunnels, highways, highway noise barriers, and urban roads. However, before tackling that discussion, we describe briefly some fundamentals from basic research that underpin this TiO₂-based photocatalytic technology. In its pristine or modified form, TiO₂ has been the most popular and most extensively investigated photocatalyst, and is the primary source of modern third generation composite photoactive materials [27,28].

Photocatalytic processes occurring in heterogeneous systems are complex and multifarious starting from the absorption of photons by the solid photocatalyst, and ending with the evolution of reaction products. This complexity is particularly reflected in the terminology used to describe various characteristics of heterogeneous photocatalysis, which, although it has come to some maturity in recent years, continues to undergo extensive developments through efforts of many researchers from the fields of catalysis, photochemistry and materials science, among others. Accordingly, prior to tackling the many deNOxing efforts in cleaning up the atmospheric environment, an important aspect of applied photocatalysis with metal-oxide semiconductor photocatalysts, it is imperative to appreciate and understand some of the fundamentals underlying Heterogeneous Photocatalysis (following Section 2 and Section 3)—the primary approach in these efforts.

2. Some Fundamentals of Heterogeneous Photocatalysis

Historically, Heterogeneous Photocatalysis is an interdisciplinary field at the intersection of Chemistry and Physics. It rests on four basic pillars (Figure 5): (a) heterogeneous catalysis; (b) photochemistry; (c) molecular/solid-state spectroscopy; and (d) materials science of semiconductor photocatalysts of interest (in the present context: metal oxides). Accordingly, it is worth looking into each of these pillars to assess how they have shaped present-day knowhow, particularly Pillars (a)–(c) (Section 2) together with the photophysics of metal-oxide semiconductors (Section 3).

Currently, photocatalysis is best described as a change in the rate of a chemical reaction or its initiation under the action of ultraviolet, visible, or infrared radiation in the presence of a substance that absorbs light and is involved in the chemical transformation of the reaction partners; the photocatalyst is the substance that causes, by absorption of ultraviolet, visible, or infrared radiation, the chemical transformation of the reaction partners, repeatedly coming into intermediate chemical interactions with them and regenerating its chemical composition after each cycle of such interactions [29].

2.1. Influence of Catalysis on Photocatalysis—Comparisons and Contrasts

Some researchers consider photocatalysis a component of the field of Catalysis. However, some postulates that are typical of traditional catalysis are somewhat antagonistic to photocatalysis. For instance, by analogy with catalysis, one of the postulates would suggest that a photocatalytic process should favor only thermodynamically allowed chemical reactions. From this point of view, light should cause a decrease of the potential energy barrier relative to the dark catalytic reaction and thus accelerate the establishment of chemical equilibrium between reagents and products in the heterogeneous system. Photoreactions that take place in heterogeneous systems and are thermodynamically unfavorable (e.g., photolysis of water) are classified as non-photocatalytic but photosynthetic, despite similarities in all major steps of the photoprocesses. The sole reason for both types of photoreactions is the free energy of the actinic light. In fact, the action of the free energy of light turns the system into a thermodynamically open system so that, by definition, the concept of thermodynamic equilibrium is not applicable.

The definitions given to photocatalysis and photocatalysts are very similar to the definitions of catalysis and catalyst commonly used in conventional (thermal) catalysis. This similarity suggests a strong impact of catalysis to the field of photocatalysis. According to current thinking, photocatalysis is viewed as an alteration of the reaction rate in the presence of a substance that interacts repeatedly with reagents subsequent to which its original state is restored after each reaction cycle just like in catalytic processes. Thus, photocatalysis could, in principle, be considered as catalysis involving the action of light. Formally speaking then, the transformation of a reagent in a catalytic process may be exemplified by a simple chemical reaction (e.g., Equation (6)):

Cat + R → Cat + P

(6)

where Cat is the catalyst, R is the reagent, and P is the product. For a photocatalytic process, we need only involve the interaction of the system with light (Equation (7)):

Cat + R + hν → Cat + P

(7)

Certain similarities between catalysis and photocatalysis support such considerations, as both are characterized by the alteration of: (i) the reaction rate; and (ii) the reaction pathway; as well as (iii) by the essential role of adsorption of the reagent R and desorption of the product P in the (photo)catalytic cycle. In addition, the red-shift of the spectral limit of a photocatalytic process, in contrast to a catalyzed photochemical process, can be taken as an analog of the decrease of the activation energy required for the catalyzed reaction compared to the non-catalyzed reaction. Not surprisingly then, some terminology, major characteristics, and quantitative parameters typically used in catalysis have been adapted to the field of photocatalysis. For example, the terms activity and selectivity of a photocatalyst, as well as the terms surface-active center, turnover number and turnover frequency are examples transferred from catalysis to photocatalysis. Within the present context, the most important feature in photocatalysis borrowed from catalysis is the notion of a surface-active center that can best be described as a surface regular site or a surface defect site capable of initiating chemical transformations of the reactants [29].

In heterogeneous photocatalysis, surface-active centers (sites) initiate a chemical sequence subsequent to the centers being activated by absorption of (light) photons by the photocatalyst. That is, before the reaction cycle, the surface-active centers (S) on the photocatalyst are initially in their inactive ground state unable to react with other molecules. However, once the solid photocatalyst has been electronically excited by absorption of photons with the appearance of free charge carriers, electrons (e⁻) and holes (h⁺), or excitons (e°), this excitation energy may be localized on the surface-active centers to yield centers in their chemically active excited states (S⁺ and S⁻; Equation (8)), which can also form by interaction with the excitons (Equation (9)), or otherwise the photons may excite the surface-active sites directly to give S* (Equation (10)). The surface-active centers in their electronically activated excited state are then capable of initiating surface chemical reactions.

S + e⁻ → S⁻

(8a)

S + h⁺ → S⁺

(8b)

S + e° → S⁻ + h⁺

(9a)

S + e° → S⁺ + e⁻

(9b)

S + hν → S*

(10)

The photoactivated states of the surface-active centers can decay back to the ground state of the centers through different physical relaxation pathways: (a) by recombination of the charge carriers trapped at surface-active centers with free charge carriers of the opposite sign, or with free excitons (Equations (11) and (12)); (b) by thermal ionization when the activation energy of ionization is comparable to kT (Equation (13)); (c) photo-ionization in the spectral range of photexcitation corresponding to the absorption band of the surface-active centers (Equation (14)); and (d) by spontaneous deactivation (Equation (15)).

S⁺ + e⁻ → S

(11a)

S⁻ + h⁺ → S

(11b)

S⁺ + e° → S + h⁺

(12a)

S⁻ + e° → S + e⁻

(12b)

S⁺ → S + h⁺

(13a)

S⁻ → S + e⁻

(13b)

S⁺ + hν → S + h⁺

(14a)

S⁻ + hν → S + e⁻

(14b)

S* → S + hν’

(15)

Concurrently, chemical reactions of molecules with the photoactivated states of the surface-active centers (S*) also lead to the return of the centers back to their ground state (S; Equation (16)).

S* + R → S + P

(16)

In summary, there are two states of surface-active centers in heterogeneous photocatalysis: (i) the initial inactive ground state S; and (ii) the chemically active excited state(s) (S⁺, S⁻ or S*). Both physical relaxation and chemical interactions of these excited state(s) with molecules return the centers back to their initial ground state.

2.1.1. Is the Process Photocatalytic or Stoichiometric?

The photocatalytic cycle begins by absorption of photons causing the activation of surface-active centers, followed by chemical reactions that terminate by desorption of product(s) from the active centers, and restoring the centers back to their initial ground state. This is essential in determining quantitatively the parameters borrowed from catalysis: namely, the turnover frequency (TOF) and the turnover number (TON) [29]. TOF refers to the number of photoinduced transformations (product formed or reactant consumed) per catalytic site per unit time as expressed by Equation (17); N_a is the number of catalyst active sites, and N is the number of photocatalytic transformations.

$$TOF=\frac{1}{N_a}\frac{dN}{dt}$$

(17)

It is important to recognize that TOF considers the number of surface-active centers in their initial ground state before photactivation. When the number of such active sites is unknown, the surface area is often used to normalize the number of turnovers—this is known as the areal turnover frequency [29].

In catalysis, TOF describes the activity of the catalytic centers. In photocatalysis, however, the rate of a chemical transformation of a molecule in a photocatalytic process depends on the light intensity, so that TOF is also light intensity-dependent and is generally taken as a characteristic feature of a photocatalyst. TOF can be used to compare the activities of various photocatalysts only if the photocatalysts were photoactivated under identical conditions. This is extremely important! Otherwise, any comparison of the activities of various photocatalysts will have no physical/chemical meaning.

Another turnover quantity taken from conventional catalysis is TON that describes the number of times an overall reaction (the photochemical transformation) goes through a photocatalytic cycle for a given period, t (Equation (18)). Again, it is important to recognize that it is the number of photocatalytic centers in their ground state that is relevant when assessing TON in photocatalysis.

$$TON=\frac{1}{N_a}{\int }_0^t\frac{dN}{dt}dt=\frac{N}{N_a}$$

(18)

Since the photocatalytic reaction rate, and thus the number of photoinduced chemical transformations, depends on light intensity, TON cannot be considered an intrinsic property of a photocatalyst. Nonetheless, TON is an important parameter in experimental studies of heterogeneous photochemical reactions as it shows whether the photochemical transformation is catalytic. Where TON > 1, the active center is restored back to its initial ground state after the chemical transformation cycle and is ready to initiate subsequent cycles: the photochemical process is then said to be photocatalytic. When TON ≤ 1 after the first cycle, the active center is completely deactivated and cannot initiate subsequent cycles of the chemical transformations; this means that the surface-active center only interacts with one molecule. In this case, the process is a stoichiometric process, as encountered in photostimulated adsorptions of molecules on solid surfaces.

If a stoichiometric surface reaction were to occur on the same surface-active center as a photocatalytic reaction, it would provide a method with which to account for the number of surface-active centers and thus aid in establishing the magnitude of TON. For instance, if a photostimulated adsorption of a molecule is the first step in a complex photocatalytic process that concludes the photocatalytic cycle at the photoadsorption step, then it is possible to determine the number of photoadsorbed molecules that corresponds to the number of surface-active centers involved in the photocatalytic reaction. This method was used by Emeline and coworkers [30] to assess quantitatively TON for the photooxidation of hydrogen by oxygen and the photoreduction of oxygen in the presence of hydrogen over solid particles of ZrO₂. The authors showed that the photostimulated adsorption of oxygen on zirconia, which by default is a stoichiometric reaction, is the first step in the photooxidation of hydrogen. Therefore, the maximal number of photoadsorbed oxygen molecules gives the number of surface-active centers for the photooxidation of hydrogen. They determined that TON > 1 for this process, and consequently the process was deemed to be a photocatalytic process.

2.1.2. Mechanistic Implications

Another common feature between catalytic and photocatalytic processes are the two general mechanisms of surface chemical reactions: (a) the Eley–Rideal mechanism (ER); and (b) the Langmuir–Hinshelwood (LH) mechanism. The Eley–Rideal mechanism assumes that molecules in the bulk solution phase interact with surface species (either pre-adsorbed molecules or surface-active centers), whereas the Langmuir–Hinshelwood mechanism presupposes that the reaction involves pre-adsorbed molecules and surface species (either pre-adsorbed molecules or surface-active centers). For decades, the LH mechanism was the favorite mechanism among experimentalists (although erroneously) because of the experimentally observed dependence of the rate of the photocatalytic process on reagent concentration (Equation (19)).

$$\frac{dC}{dt}=\frac{kKC}{1+KC}$$

(19)

where k is the apparent rate constant, K is the apparent equilibrium constant, and C is the concentration of reactant molecules. This equation correlates with the Langmuir isotherm (Equation (20)):

$$\theta =\frac{K_{L}C}{1+K_{L}C}$$

(20)

where K_L is the adsorption/desorption equilibrium constant (K_L = k_ads/k_des) provided the rate of the process is proportional to the surface coverage (θ) by pre-adsorbed molecules (Equation (21)). That is, the adsorption/desorption equilibrium is not disturbed during the photoprocess.

$$\frac{dC}{dt}=k\theta$$

(21)

If K = K_L, then by default the observation that the rate dependence followed Equation (19) was taken as an indicator that the process followed the LH mechanism. However, for many heterogeneous photoprocesses, it was established conclusively that K ≠ K_L and K depended on light intensity. Later, Emeline and coworkers [31,32] demonstrated that the dependence expressed by Equation (19) was also consistent with the ER mechanism, considering that the lifetime of the activated state of the surface-active centers is limited (see above). Moreover, if the physical relaxation pathway of the excited state of the active center involved recombination or photo-ionization, then K became dependent on light intensity in accord with experimental results (Figure 6 and Figure 7).

The approach based on a limited lifetime of the excited state of active centers is applicable to both ER and LH mechanisms, provided a quasi-steady-state is valid for the concentration of active centers in the activated state. The alternative view (i.e., LH mechanism) is also based on a quasi-steady-state approach for the concentration of adsorbed molecules and on a strong disruption of the adsorption equilibrium in accord with experimental data [33,34]. Kinetic measurements alone, as typically conducted in photocatalytic studies, cannot distinguish unambiguously which mechanism is operative in a photocatalytic process. Whatever the actual mechanism (ER or LH), however, the role of adsorption of a molecule on the surface is not insignificant and in fact is typical in heterogeneous photocatalysis. The only difference occurs in the ER mechanism, whereby photostimulated adsorption plays an important role rather than dark adsorption, with the latter so crucial to the LH mechanism.

In summary, heterogeneous catalysis has had a non-insignificant impact on the field of heterogeneous photocatalysis. Nonetheless, a serious discrepancy between these two phenomena remains. In catalysis, the catalyst is in its electronic ground state and is in thermodynamic equilibrium with the environment at a given temperature; it accelerates both direct and back reactions, and promotes a faster establishment of the reaction equilibrium. By contrast, in heterogeneous photocatalysis, the photocatalyst becomes active in an electronically excited state that is not in thermodynamic equilibrium with its surroundings, so that the sole relaxation pathway back to its ground state is acceleration of the direct reaction, because the photocatalytic heterogeneous system is an open system under a permanent flow of light energy as occurs in photochemistry.

2.2. Influence of Photochemistry on Photocatalysis—Comparisons and Contrasts

Photochemistry of heterogeneous systems is also an important pillar of photocatalysis, as it considers thermodynamically open systems. Thus, photocatalysis can also induce reactions that lead to light energy conversion and storage through formation of higher energy products, as occurs in natural photosynthesis [35,36]. Concepts taken from photochemistry suggest that the photocatalytic process occurs through electronically excited states of the photocatalyst prompted by light absorption. It should be noted that photocatalyzed reactions are often indistinguishable from photosensitized reactions examined in photochemistry [37]. Historically, many photocatalytic reactions were considered photosensitized reactions, so that what was once called a photosensitizer is now referred to as a photocatalyst. Two major characteristics of photosensitized reactions are also typical of photocatalytic processes: (i) a red-shift of the spectral limit of the photoreaction; and (ii) the alteration of the reaction pathway compared to that of the photochemical reaction. The red-shift of the spectral limit of a photoprocess and alteration of the reaction pathway are distinguishable fingerprints of heterogeneous photocatalytic reactions. Enhancing the ability of photocatalysts to sensitize photochemical reactions, especially toward visible light, is a major challenge in applied heterogeneous photocatalysis.

Photochemistry is concerned with the chemical effects of ultraviolet, visible, or infrared radiation while the photocatalyst is a substance that can produce on absorption of light chemical transformations of the reaction partners subsequent to formation of an excited state of the photocatalyst, which interacts repeatedly with the reaction partners forming reaction intermediates and regenerates itself after each cycle of such interactions [37]. Hence, a photocatalytic process (Equation (22)) is similar to a photochemical reaction (Equation (23)).

Cat + hν → Cat* + R

(22)

R + hν → R* → P

(23)

where Cat and Cat* denote the photocatalyst in the ground state and in its electronically excited states, respectively; R is the reagent in its ground electronic state; R* is the reagent in the electronically excited state formed by absorption of photons; and P is the product of the photochemical/photocatalytic transformation.

Thus, a photocatalytic process could be viewed as a particular case of a photochemical reaction, whereby the photocatalyst plays the double role as one of the reagents and as one of the products subsequent to restoring its initial ground state.

2.2.1. Photocatalytic Versus Photochemical Processes

There are far more mechanistic similarities between photocatalytic and photochemical reactions than there are between photocatalyzed and catalyzed reactions. Both the photochemical reaction and the photocatalytic reaction require absorption of photons to form electronically excited states of one of the reaction partners, which cause distortions of the initial thermodynamic equilibrium and both reactions then occur through non-equilibrium states. This means that the subsequent sequence of molecular transformations can only proceed in one direction to restore the equilibrium state of the system and there is no path for a back reaction. Both photoreactions are characterized by the existence of physical relaxation pathways of the electronically excited states: radiative and non-radiative relaxation processes. For a solid photocatalyst, these relaxation processes are excitonic decay, free charge carrier recombination, and charge carrier trapping. In addition, all photocatalytic processes obey the general law of photochemistry: only absorbed photons initiate chemical transformations, and one absorbed photon results in only one elementary transformation. Accordingly, photocatalysis can be viewed as catalysis of a photochemical reaction. Consequently, three major characteristics used in photocatalysis have been adopted from photochemistry: (a) spectral sensitivity of a photoprocess; (b) photoactivity of a photocatalyst; and (c) selectivity of a photocatalyst.

Spectral Sensitivity of a Photoprocess

The spectral sensitivity of a photoprocess is described by the spectral range wherein photon absorption by the system initiates a chemical transformation; for a photocatalytic process this is the red spectral limit that corresponds to the lowest energy photons to initiate photocatalytic cycles. Note that the blue spectral limit of all photocatalysts expands into the vacuum UV region and so cannot be determined under typical conditions. We have often noted that the photocatalysis literature claims (albeit incorrectly) that photons with energy lower than bandgap of the photocatalyst cannot initiate chemical reactions, and so the red spectral limit of a photocatalytic process is taken as the bandgap energy. Theoretically, this is true only for ideal solids with no defect states, although even for an ideal solid structure there are always regular surface states (e.g., the Tamm and Shockley states) whose energies of photoexcitation could be lower than the bandgap energy. In practice, however, one deals with real solid photocatalysts whose structures possess various imperfections: for example, intrinsic point defects, impurity defects, dislocations, and defect surface states (e.g., add-atoms, corners, edges, steps etc.). Photoexcitation of such defect states results in a significant shift of the red limit of the photocatalytic effect toward lower energy photons compared to the bandgap energy—for the case of TiO₂, although the bandgap energies are 3.0 eV for the rutile polymorph and 3.2 eV for the anatase counterpart, the red limits for both oxidative and reductive pathways are 2.2 eV (λ = 560 nm), well below bandgap.

Accordingly, from a practical application of heterogeneous photocatalysis, the longer is the wavelength of the red spectral limit of the photocatalytic process, the larger is the fraction of sunlight that can be exploited for a higher overall activity of the photocatalytic system. Increasing the spectral sensitivity of a photocatalyst toward visible light is a major challenge in applied photocatalysis, currently being examined through modification of pristine photocatalysts by metal and non-metal doping and by physical modifications to form intrinsic defects, which give rise to extended extrinsic absorption of visible light. A detailed assessment of this issue and corresponding problems have been described elsewhere [38,39,40].

Photoactivity of a Photocatalyst: The Quantum Yield Φ

The photoactivity of photocatalysts describes the ability of a solid to transform the absorbed actinic light into a chemical sequence, for which the quantum yield (Φ) is its most relevant parameter. In heterogeneous photocatalysis, Φ is the ratio between the number of molecules that have reacted (N_m) to the number of photons actually absorbed (N_hν) at a given wavelength of the actinic light (Equation (24)), provided the photocatalytic process has reached a stationary state [29], otherwise Φ can be expressed by the differential form (Equation (25)):

$$\Phi =\frac{N_m}{N_{h\nu }}$$

(24)

$$\Phi =\frac{\frac{d{N}_m}{dt}}{\frac{d{N}_{h\nu }}{dt}}$$

(25)

Typically, Equation (25) is used since the reaction rate, and thus Φ, depends on such parameters as time, temperature, wavelength of the actinic light, concentration of reagent, light intensity, and the solution pH (among others). Regrettably, the rate of absorption of photons and the number of absorbed photons remain inaccessible quantities in heterogeneous photocatalysis because of experimental limitations. Consequently, the photonic yield (ξ) was introduced and defined in a manner similar to the quantum yield (Φ) with the main difference being that reference is made to the number of photons of a given wavelength of the actinic light (N°_hν) incident on the photoreactor, and not on the number of photons absorbed by the photocatalyst under stationary conditions, as expressed by Equation (26) or in differential form by Equation (27) [28].

$$\xi =\frac{N_m}{N_{h\nu }^o}$$

(26)

$$\xi =\frac{\frac{d{N}_m}{dt}}{\frac{d{N}_{h\nu }^o}{dt}}$$

(27)

Both Φ and ξ are defined as in photochemistry [37]; the former is more appropriate, however, since only absorbed photons can initiate interfacial chemical reactions, thus Φ is an intrinsic characteristic of a photocatalyst. Nonetheless, ξ may be a more practical parameter to assess experimentally as it depends on the number of incident photons and not on the number of photons actually absorbed by the photocatalyst; ξ is also an intrinsic characteristic of a photocatalyst.

In the field of Chemical Engineering, the parameters often used are quantum efficiency and photonic efficiency; the former describes the rate of a given photophysical or photochemical process divided by the total photon flux absorbed [29]; it applies especially when using polychromatic radiation to activate the photocatalyst. The photonic efficiency describes the ratio of the rate of the photoreaction measured at t = 0 (initial rates) to the rate of incident photons within a given wavelength range [29]. Consequently, these parameters provide an estimate of the overall photoactivity of the photocatalysts, but provide no information regarding light absorption, photoexcitation mechanisms, and specific details as to the efficiencies of excitation transfer to the surface or to the initiation of surface chemical transformation as they denote engineering efficiencies.

We cannot overemphasize that the above parameters used in photocatalysis and borrowed from photochemistry characterize the efficiency of photocatalysts, unlike TON and TOF from catalysis which are light intensity-dependent. Photochemical parameters may also depend on the intensity of photoexcitation; however, experimental conditions can be controlled such as to obtain a linear dependence of the reaction rate on light intensity as established by Emeline and coworkers [30,31] to obtain the maximal photoefficiency of photocatalysts when all the quantum and photonic parameters become independent of light intensity. The quantum yield value (and other photochemical parameters) is essentially governed by the reaction rate of a heterogeneous photocatalytic reaction that depends on the stationary surface concentration of charge carriers (either electrons or holes) (Equations (28) and (29)).

R + e⁻ → P

(28a)

R + h⁺ → P

(28b)

d[R]/dt = k [R] [e]

(29a)

d[R]/dt = k [R] [h]

(29b)

Selectivity of a Photocatalyst

The selectivity of a photocatalyst is the ability of a photocatalyst to drive a photoprocess toward a certain reaction product. Emeline and coworkers [41,42,43,44] established two possible causes for the selectivity of a photocatalyst when the rate of the surface photochemical reaction is dictated by the surface concentration of photocarriers (electrons for reduction, holes for oxidation).

The first originates from the ratio between the concentrations of electrons and holes at the surface of a photoactive material, so that alteration of this ratio caused by various factors leads to changes in the ratio between the efficiencies of surface redox reactions for molecules displaying both electron-acceptor and electron-donor behavior. For example, if a given reaction product (P_i) were formed by a reduction pathway during the photostimulated surface reaction with reagent (R) interacting with both electrons (e) and holes (h), the surface selectivity toward the formation of the given product (S_Pi) would then be expressed by Equation (30): k_e,i and k_h,j are the rate constants for a given reaction pathway. A similar expression can be formulated for the products formed by the oxidation reaction pathway.

$$S_{P_i}=\frac{\frac{d[P_i]}{dt}}{\frac{d\left[R\right]}{dt}}=\frac{k_{e,i}\left[e\right]}{{\sum }_i{k}_{e,i}\left[e\right]+{\sum }_j{k}_{h,j}\left[h\right]}$$

(30)

By default, it is assumed that the reaction rate does not depend on reagent concentration, which would correspond to saturation of the rate dependence on reagent concentration for the LH-like kinetics. Equation (30) clearly indicates that the selectivity of the photocatalyst is determined by the ratio between the surface concentrations of electrons and holes.

The second origin of selectivity is related to the formation of surface-active sites of a different type, which may display different activity and selectivity. Different types of surface-active sites can be formed through photoexcitation of localized surface and sub-surface electronic states (intrinsic defects and/or dopant states), or through formation on different types of surface structures (e.g., dominating surface plane). This type of selectivity is clearly observed for molecules that demonstrate only electron-donor or electron-acceptor behavior, although it is typical of any surface photochemical reaction. As such, the selectivity of the surface toward formation of a given reaction product can be expressed by Equation (31), which is determined by the ratio of a given rate constant characteristic of a given sort of active site to the sum of rate constants corresponding to all possible types of active sites on the surface of the photocatalyst (here rate constants are apparent constants and implicitly include the surface concentrations of active sites). Obviously, this factor (ratio of rate constants) is also significant for the selectivity in the earlier scenario (Equation (30)).

$$S_{P_i}=\frac{\frac{d[P_i]}{dt}}{\frac{d\left[R\right]}{dt}}=\frac{k_{e,i}}{{\sum }_i{k}_{e,i}}$$

(31)

Unlike macro crystals with their clearly manifested crystal faces, the dominant surface planes in nanoparticles of photocatalysts are strongly affected by the method and conditions of their synthesis. Concomitantly, in the course of photoprocesses, various planes of photocatalyst particles may display different selectivities, which may be due to different atomic structures, or magnitude and distribution of surface charge, or to a dominant type of defects (e.g., surface-active centers, among others) [45,46,47,48,49]. Hence, altering the nanoparticle surface structure via formation of various dominating faces with strong selectivity toward either oxidation or reduction can alter the overall selectivity of the photocatalyst.

2.2.2. Correlation between Activity and Selectivity

The correlation between activity and selectivity of a photocatalyst has been demonstrated by Emeline et al. [50] in the photodegradation of 4-chlorophenol taking place over irradiated TiO₂ with formation of three primary intermediates: (i) benzoquinone formed by a reductive pathway; (ii) 4-chlorocatechol formed by an oxidative pathway; and (iii) hydroquinone produced by both oxidative and reductive pathways. Figure 8 shows that the selectivity toward formation of hydroquinone and the activities of the photocatalyst are strongly correlated (r = 0.984; p = 0.0004).

According to the balance of charge, as expressed by Gerisher [51], a true (photo)catalytic process is characterized by the equality of the rates of consumption of electrons and consumption of holes in the overall reaction (Equation (32)). That is,

$$\frac{d\left[e\right]}{dt}=\frac{d\left[h\right]}{dt}$$

(32)

This charge balance is a required condition for the effective photocatalytic process, or else the deviation from catalytic equilibrium (Equation (32)) would result in the transformation of charge balance according to Equation (33) [38,52].

$$\frac{d\left[e\right]}{dt}+\frac{d\left[F\right]}{dt}=\frac{d\left[h\right]}{dt}+\frac{d\left[V\right]}{dt}$$

(33)

where F and V denote, respectively, electrons and holes trapped in bulk defects (color centers), which accelerate bulk charge carrier recombination and decrease the activity of the photocatalyst. Since formation of hydroquinone consumes both electrons and holes, these reaction pathways create a favorable condition for the photodegradation of the phenol being truly photocatalytic by suppressing bulk recombination. Therefore, the higher is the selectivity of the photocatalyst surface toward formation of hydroquinone, the higher is the activity of the photocatalyst during the photodegradation. In general, this rule can be formulated thus: a higher activity of photocatalysts can be expected provided both reductive and oxidative pathways occur with equally high efficiency.

2.3. Influence of Molecular Spectroscopy on Heterogeneous Photocatalysis

Another major pillar of heterogeneous photocatalysis is molecular spectroscopy of adsorbed molecules. Most pioneering studies focused on understanding the changes in molecular structure induced by adsorption of molecules on solid surfaces [53,54,55], with light being an active factor in transforming these adsorbates. The effects of photodissociation and photodesorption of adsorbed molecules have been established together with the photoadsorption of simple molecules on dispersed semiconductors [55,56].

The photocatalytic cycle begins with the absorption of light quanta by the solid photocatalyst and culminates with the chemical transformations of surface-adsorbed molecules, ultimately evolving reaction products into either the gaseous or liquid phase. The role of the photocatalyst and corresponding photophysical events taking place in solids are often treated in a simplistic manner. The ensemble of particles that absorb photons is the light harvesting system, whereby the photocatalyst particle is both a sensitizer and the source of intermediates—i.e., photoelectrons and photoholes. In addition, the intrinsic, that is, the fundamental absorption of light by the solids is of primordial importance in photocatalysis; this was a reasonable approach in most studies that were oriented on mechanistic investigations of chemical reactions, or else on practical applications of heterogeneous photocatalysis. The complexity and variety of photophysical processes in solid photocatalysts, together with the interdependence between physical and chemical events at the microparticles’ and nanoparticles’ surfaces, must always be kept in mind, even in applied heterogeneous photocatalysis. Accordingly, next we describe some relevant events that precede surface chemical reactions on the photocatalyst particle (Figure 9).

2.3.1. Absorption of Light by Solid Photocatalysts

Absorption of electromagnetic radiation represents the transfer of energy from an electromagnetic field to a material (the photocatalyst) or to a molecular entity [29]. The transformation of light energy into the energy of electronic excitations (electrons, holes) of a solid photocatalyst is fundamental in heterogeneous photocatalysis, with these excitations being created as a result of absorption of photons in a solid during some time (lifetime of charge carriers) before relaxation (recombination of charge carriers), i.e., before transformation of the electronic energy into thermal energy. Electron–hole pairs in solids then transfer the energy within some sub-surface space, the depth of which depends on both the linear absorption coefficient of the photocatalyst and the diffusion length of the photocarriers and, in some particular cases, on particle size.

2.3.2. Quantities that Describe Light Absorption in Heterogeneous Photocatalysis

Dispersed (powdered) photocatalysts in contact with gaseous or liquid phases in irregular shaped reactors are typically the light absorbing media, and so require consideration of the absorption and scattering of light together with a few quantities that we now outline for the absorption of light used in experimental photocatalytic reactions.

Absorption of light is manifested by the decrease of the energy of the light beam when it passes through a medium because of the transformation of the photon’s energy into another energy form. Experience has taught us that I_R + I_T < I₀ for light absorbing substances and so in accord with the energy conservation law, the intensities of the incident (I₀), transmitted (I_T) and specular reflected (I_R) beams are given by Equation (34),

I₀ = I_R + I_T + I_A

(34)

where I_A is the intensity of the absorbed light beam. For moderate intensities of the incident light beam and in the absence of additional input of energy to the substance, we have (Equation (35))

I_R = R I₀

(35a)

I_T = T I₀

(35b)

I_A = A I₀

(35c)

The dimensionless coefficients R (here it denotes the reflectance), T (transmittance) and A (absorbance) in Equations (34) and (35a)–(35c) satisfy Equation (36),

R + T + A = 1

(36)

Typically, absorbance can be determined optically using a dual-beam spectrophotometric technique such that the ratio I_T/I₀ is detected, while reflectance is compensated using a reference sample. From experience, we have that (Equation (37))

$$I_T=\left(1-R\right)I_0{e}^{-\alpha d}$$

(37)

And, from Equations (35a)–(35c) and (37), we obtain for A (Equation (38)) that

$$A=\left(1-R\right)\left[1-e^{-\alpha d}\right]$$

(38)

which describes the Lambert–Bugger law for absorbance by a solid parallel plate of thickness d; the quantity α is the linear (Naperian) absorption coefficient (in cm⁻¹). In general (Equation (39)),

$$I_{\left(x\right)}=\left(1-R\right)I_0{e}^{-\alpha x}$$

(39)

where I_(x) is the intensity of the light beam a distance × from the illuminated surface of the optically uniform plate.

Equation (39) can also be obtained from Equation (40) which expresses the independence of the absorption coefficient α from the light intensity,

dI = −α I_(x) dx

(40)

where dI is the fraction of light absorbed by a thin layer dx a distance × from the plate surface. The density of absorbed photons (Equation (41)) can be deduced from Equations (39) and (40). If the light beam of intensity I₀ were given in units of photons per unit area per second, then the units of dI/dx are cm⁻³ s⁻¹.

$$\frac{dI}{dx}=\alpha I_0{e}^{-\alpha x}$$

(41)

The absorption coefficient α is a spectrally dependent parameter (α = α_(λ)), whose magnitude varies in a wide range for different solids and can reach values up to ca. 10⁶ cm⁻¹ for the fundamental absorption band but can be very small for extrinsic absorption bands of solids.

Where absorption of light is due to defects or to structural imperfections (i.e., in the extrinsic absorption region for a transparent solid), the linear absorption coefficient α is proportional to the concentration of defects n (Equation (42)); σ_a is the absorption cross section of a defect.

α = σ_a n

(42)

Equation (42) represents Beer’s law (for defects in solids); the law fails at very high defect concentrations (typically > 10¹⁸ cm⁻³) when the interaction between defects in the solid becomes significant.

Taking into account Equations (42) and (37)–(39), a comparison with corresponding formulas in UV-Vis absorption spectroscopy of solvents indicates that the wavelength-dependent quantity σ_a becomes the extinction coefficient that characterizes the optical properties of a given defect in the solid. The absorption cross-section, σ_a, for optical transitions in defects that are allowed by appropriate selection rules can reach values of ~10⁻¹⁶ cm². Consequently, α ≈ 10² cm⁻¹ for n ≈ 10¹⁷–10¹⁸ cm⁻³; this is typical of imperfect photocatalyst particles. When the nominal concentration of regular entities (atoms, ions) in solids (n ≈ 10²² cm⁻³) is multiplied by σ_a ≈ 10 ⁻¹⁶ cm² leads to a realistic estimation of the absorption coefficient (α ≈ 10⁶ cm⁻¹) for fundamental absorption bands, despite breaching the conditions of Beer’s law validity. Variation of α with wavelength of the actinic light is an important factor in the activity of solid photocatalysts and other spectrally dependent parameters.

2.3.3. Absorbance and Reflectance of Photocatalysts in Powdered Form

Powdered solids with sufficiently high BET surface areas (S_BET = 10–100 m² g⁻¹) are typically used in photocatalytic studies and correspond to characteristic mean particle sizes in the microscale to nanoscale comparable to the wavelength of the actinic light. Contrary to the optically uniform plate noted earlier, light scattering now plays a major role in the action of light on powdered photocatalysts. Because of light scattering, a powdered photocatalyst in pellet form is now opaque to light compared with the optically uniform plate of the same material, even in the spectral region where absorbance is very low (extrinsic absorption region). Consequently, most of the light is reflected back in various directions and gives rise to diffuse reflectance. The light intensity decreases with distance from the illuminated side of the pellet. In this case, the Lambert–Bugger law is not valid in its original form (Equation (6)) so that in practice the energy conservation law (Equation (34)) should be applied to powdered materials with some restraint since care must be taken in measuring the diffuse reflected light.

Diffuse reflectance spectroscopy has proven a useful method in experimental measurements of the optical absorption of powdered metal-oxide photocatalysts. For instance, the absorbance A of a sample and the diffuse reflectance coefficient, R, are related by Equation (43) for zero transmittance (i.e., for T = 0); in practice, this is valid when the thickness of the powdered sample is around 3–5 mm.

A = 1 − R

(43)

UV-Vis-NIR diffuse reflectance spectra, R_(λ), are usually recorded with spectrophotometers equipped with an integrating sphere assembly and a standard reference sample of known diffuse reflectance spectrum. Accordingly, R_(λ) or A_(λ) spectra provide the spectral information about the initial state of a powdered sample.

Additional spectral information can be obtained from difference diffuse reflectance spectra, ∆R_(λ), when changes in reflection/absorption occur from controllable treatments of the photocatalyst (e.g., annealing or illumination in a vacuum or in the presence of gases) (Equation (44)).

∆R_(λ) = R_1(λ) − R_2(λ)

(44)

where R_1(λ) and R_2(λ) are the diffuse reflectance spectra measured, respectively, before and after the treatment. The case where ∆R_(λ) > 0 corresponds to the treatment-induced absorption spectrum. Hence, from Equation (43) we obtain (Equation (45)):

∆R_(λ) = ∆A_(λ) = A₂ − A₁

(45)

Thus, alteration of the diffuse reflectance coefficient in photocatalytic experiments corresponds to changes in the absorbance of the photocatalyst sample, which for practical applications is more important than changes in the absorption coefficient (such as quantum yield measurements) as it provides the information about the spectral behavior of the whole heterogeneous system. Difference diffuse reflectance spectra (∆R_(λ)) not only increase significantly the accuracy of measurements but make information accessible with regard to absorption spectral shapes induced by whatever treatment the photocatalyst is subjected to. Hence, measurements of a set of induced absorption spectra of different spectral shapes allow for a numerical analysis of the spectral data that could be very important for absorption spectra consisting of several overlapping single absorption bands.

2.4. Intrinsic and Extrinsic Absorption of Light by Semiconductor Photocatalysts

Two different spectral regions of light absorption can be distinguished in the field of optics and in the photophysics of the solid state: (i) intrinsic or fundamental absorption; and (ii) extrinsic absorption of light. Intrinsic light absorption is due to photoinduced electronic transitions between occupied delocalized states in the valence band and unoccupied delocalized states in the conduction band of semiconductor photocatalysts [57,58].

The intensity of the inter-band photoexcited transitions, characterized by the coefficients α and σ_a, is determined by selection rules similar to those for atomic photoexcitation and by the additional requirement for a quasi-momentum conservation. As a case in point, photoinduced electronic transitions in TiO₂ from the top of the valence band (mostly oxygen p-states) to the bottom of the conduction band (mostly titanium d-states) are allowed transitions (p → d transitions) as are the p → s transitions in alkali halides. The intensities of photoinduced transitions from the VB to the CB in alkali halides are much greater than those observed in TiO₂. The reason for this difference in solids may be found in the momentum conservation law for electrons that requires the fulfillment of the condition expressed by Equation (46):

k₁ − k₂ = k_hν

(46)

where k₁ and k₂ are the wave vectors of electrons in the states between which the transition occurs and k_hν is the wave vector of the photon which, when absorbed by the solid, causes the transition. To the extent that the momentum of photons resulting in the electronic transition in a typical spectral range of photoexcitation of a solid is much smaller than the quasi-momentum of electrons occupying the band states, the condition expressed by Equation (44) can be simplified to Equation (47) with the proviso that k_hν ≈ 0.

Δk = k₁ − k₂ ≈ 0

(47)

The transition between states in the VB and CB bands that follows Equation (46) is referred to as a direct transition (i.e., the transition preserves the momentum of the electron) and is characterized by high intensity at the edge of the fundamental absorption. Where the positions of the maxima do not coincide in k-space (typical of TiO₂), direct transition between these states is then forbidden because momentum conservation is not preserved. Nevertheless, photoinduced electronic transitions between such states can occur via involvement of phonons that possess the required momentum. Accordingly,

Δk ± k_hω ≈ 0

(48)

where k_hω is the wave vector of either the absorbed or the emitted phonon. Such transition is referred to as an indirect transition characterized by lower intensity as it requires a three-body interaction of the electron, the photon and the phonon.

Extrinsic light absorption in solids may originate from the photoexcitation of such defect states as: (i) zero-dimensional (0-D) intrinsic point defects (e.g., vacancies, interstitials) and impurity atoms/ions; (ii) one-dimensional (1-D) linear defects (e.g., dislocations); and (iii) two-dimensional (2-D) states (e.g., intrinsic surface states).

Special interest in heterogeneous photocatalysis concerns the surface absorption of light associated with the electronic excitation of surface states, for which the energy of absorption corresponds to the extrinsic spectral region of the solids. In this regard, the greater the specific surface area is, the greater is the impact of surface absorption into extrinsic absorption. The important issue here is that the generation of charge carriers that may induce surface chemical reactions occurs only at the surface of the solid photocatalyst.

To recap, the various absorption bands corresponding to photoexcitation of different types of defects, impurities, and surface states can cover a wide spectral range of extrinsic light absorption. This photoexcitation generates free charge carriers or surface-localized excited states that can initiate interfacial chemical processes. Accordingly, the red spectral limit of a heterogeneous photochemical or photocatalytic reaction corresponds to photon energy much less than the bandgap energy of the solid photocatalysts. The photoactivity in the extrinsic absorption spectral region is typical of most metal-oxide photocatalysts, including TiO₂.

3. Photophysical Processes in Metal-Oxide Photocatalysts

3.1. Intrinsic Structural Point Defects in Metal Oxides

Point defects related to anion (oxygen) and cation (metal) vacancies in sub-lattices are the main types of defects in metal-oxide semiconductors/insulators. Structure, together with the optical and EPR properties of families of both oxygen vacancies (V_o) and cation vacancies (V_m) are now fairly understood for wide bandgap metal oxides such as MgO, Al₂O₃, and ZrO₂ (among others) [59,60,61,62,63,64,65,66,67].

3.1.1. Defects Related to Oxygen Vacancies (V_o)

There are three main types of defects in the oxygen sub-lattice of MgO related to oxygen vacancies, the so-called F-type centers: (i) doubly charged (with respect to the lattice) and optically silent anion oxygen vacancies (F²⁺ centers); (ii) anion vacancies with one trapped electron (F⁺ centers); and (iii) two trapped electrons (F centers) [59,60,61,62]. A main feature of F⁺ and F centers is a strong localization of the electron(s) within the vacancy. Semi-empirical quantum chemical calculations based on the INDO approach [61] have shown that the optimized geometry of oxygen vacancies with both bare and trapped electrons is characterized by a displacement of the nearest-neighbor cations in a direction away from the vacancy. Atomic relaxation is largest for F²⁺ centers (6.5% of regular Mg–O distance), whereas cation displacement for neutral F centers is <2% [61]. Calculations of ionic MgO crystals have shown that electrons trapped by V_o are indeed localized in the vacancy, and that the effective charge of F⁺ and F centers in the ground state was −1.002 and −2.002, respectively, while the changes in the effective charges of Mg and O ions surrounding V_o were inconsequential with respect to regular ones [61].

In the wurtzite ZnO lattice, the value and direction of the relaxations of Zn atoms adjacent to an oxygen vacancy strongly depend on the charge state of the defect. For the F center (V_o^o), the Zn atoms are displaced inward by approximately 11–12%, while, for the F²⁺ center (V_o²⁺), an outward relaxation of about 19–23% occurs (Figure 10) [64].

The optical absorption by F⁺ and F centers and electron processes subsequent to absorption of light is of some importance in metal oxides, because the energy of a photostimulated electron transition from the ground state to the exited state and the position of the exited state within the bandgap relative to the bottom of the CB band determine the formation of free electrons in the CB band.

An explanation of the formation of free charge carriers by photoexcitation of F or F⁺ centers had been proposed in two earlier studies by Kuznetsov and coworkers [68,69]; it is illustrated schematically in Figure 11. It was implied that the first photophysical event is the optically- activated electronic transition from the ground state F (or F⁺) center to its excited state (F)* (or (F⁺)*). Free electrons in the conduction band then result from a thermally-activated electron transition from the F* (or (F⁺)*) state to the conduction band, while the free holes in the valence band appear through a thermally-activated capture of an electron from the valence band by the lower level of the excited F* (or (F⁺)*) center. After the photoformation of electron–hole pairs through such a scheme, the F* (or (F⁺)*) center returns to its initial ground state that ensures the stability of these processes during the photoexcitation events. The thermal energy needed for the occurrence of such processes is about 0.1 eV for the 2.95 eV absorption band and ~0.5 eV for the 2.56 eV band that significantly exceeded the energy of phonons at the temperatures employed (bandgap energy E_g of VLA rutile was 3.05 eV at 90 K and 3.01 eV at 290 K). A recent study based on experimental results and literature data further led Kuznetsov and coworkers [70] to hypothesize that, following the prime optical excitation of defects in the solid (Ti³⁺ centers), the heat released during the non-radiative electron transitions dissipated into the nearest neighborhood of these centers with consequences (see Section 5). Localized non-equilibrated excitation of the phonon subsystem was equivalent to energies up to 1 eV.

Experimentally determined positions of the absorption band maxima of F and F⁺ centers in several other metal oxides have been reported [60,61,71,72]. In accord with experimental results, theoretical calculations placed the ground state levels of both F and F⁺ centers in MgO (bandgap, 7.8 eV) at 3 eV above the top of the VB band [61]; this means that the exited states of these centers are near the bottom of the CB band (experimentally, ca. 0.06 eV below CB; see Ref. [61] and references therein). In the case of TiO₂, the F centers have been located at 2.9 and 2.55 eV [61].

The ground state of the F center in α-Al₂O₃ is at 5.3 eV from the VB band, and thus the excited state F* lies within the conduction band manifold [61]. The ground state of the F⁺ center is distant from the VB band by 3.1 eV so that the exited state (F⁺)*, corresponding to the absorption band at 6.3 eV, lies within the CB band. Indeed, irradiation at 6.1 eV within the spectral range corresponding to the F⁺ center’s absorption band produces photoconductivity down to at least 10 K (see Ref. [63] and references therein). For other F⁺-center’s exited states corresponding to the 5.4 eV and 4.8 eV absorption bands are 1.0 eV and 1.7 eV distant, respectively, from the CB band. Thus, UV irradiation in the absorption bands of F and F⁺ centers in MgO or F centers in α-Al₂O₃ result in the photo-ionization of the defects and ultimately initiation of surface reactions in the absence of any significant number of bulk electron traps.

Absorption spectra of strongly defective crystals of MgO and Al₂O₃ revealed additional absorption bands that were ascribed to aggregates of oxygen vacancies [61,73,74,75,76], the main types being F₂ centers (i.e., 2 V_os with 4 trapped electrons), and F₂⁺ and F₂²⁺ centers (two V_os with three and two trapped electrons, respectively). Under irradiation with fast neutrons (E > 1.2 MeV) [75] or bombardment by Cu⁺ or Ti⁺ ions (E = 30 KeV) [74], F and F⁺ centers were shown to be the dominating defects in the oxygen sub-lattice. It is evident that the family of intrinsic defects related to oxygen vacancies yields in toto broad absorption spectra covering the UV, visible and near-IR spectral regions.

A full understanding of the pathways of photoactivation of wide bandgap metal oxides requires a detailed examination of the photoconversion F⁺ → F. In his 1982 review article, Crawford called attention to a puzzling feature of the behavior of F⁺ centers in Al₂O₃ when irradiated into the 4.8 eV and 5.4 eV absorption bands, which decreased the number of F⁺ centers and increased the number of F centers (absorption band at 6.0 eV) [63]. Photoconversions F⁺ → F were also found in MgO crystals and were accompanied by formation of hole centers [61,62,63], which excludes thermal ionization of the excited (F⁺)* center followed by further electron trapping by another F⁺ center. Hole formation was connected with the spontaneous electron capture from an O²⁻ adjacent to the (F⁺)*-center’s empty ground state [63].

Semi-empirical calculations have also offered explanations for the F⁺ → F photoconversions [61,62]. Theory predicted the existence of several quasi-local energy levels of the F⁺ center in the upper part of the VB band, which consist mainly of atomic orbitals of O²⁻ ions surrounding the F⁺ center. Under optical excitation, electron transition from these quasi-local levels to the ground state of the F⁺ center yielded an F center and a hole in the VB band [61,62]; ultimately, the hole was trapped either by a cation vacancy V_m or by some impurity.

Predictable schemes of the photogeneration of electrons and holes on excitation of F-type centers in wide bandgap oxides can be summarized by Equations (49)–(51c).

F + hν → F* → F⁺ + e⁻_cb

(49)

F⁺ + hν → F⁺* → F²⁺ + e⁻_cb

(50)

F⁺ + hν → F⁺*

(51a)

F⁺* + (O²⁻)_vb → F + (O⁻)_vb

(51b)

(F⁺)_ql + hν + (O²⁻)_vb → F + (O⁻)_vb

(51c)

Equations (49) and (50) represent cases of the photoionization of defect centers with formation of free electrons in the CB band. Equation (51b) corresponds to electron capture from an O²⁻ ion adjacent to the F⁺*-center, whereas Equation (51c) represents electron excitation from quasi-local levels of an F⁺-center (i.e., F⁺_ql) in the VB band accompanied by formation of an F center.

Emeline and coworkers [77] reported a single maximum at 4.6 eV in the spectral dependence of the quantum yield of photoadsorption of O₂ on powdered MgO, whereas Kuznetsov et al. [78] and Lisachenko [79] reported the maximum to occur at 5.0 eV. Spectral dependences of Φ of photo-adsorption of H₂ and CH₄ on powdered MgO displayed maxima at 4.6–4.55 eV [77] and a well-resolved shoulder at 3.75 eV. The photogeneration of holes on excitation within the spectral range, corresponding to the absorption band of F⁺ centers with maximum at 4.9 eV, was likely responsible for the photoadsorption; however, the spectral feature at 3.75 eV found no counterpart in the absorption spectra of F⁺ and F centers. Only the absorption band of an F₂ center at 3.63 eV was near the feature at 3.75 eV, but nothing is known about the processes that followed the photon absorption by this center. Note that hole trap point defects related to cation vacancies in MgO absorbed light in the region 2.2–2.35 eV [67], far from the spectral feature just described.

Spectral efficiencies of the photoadsorption of O₂ and the photodissociation of N₂O on Al₂O₃ displayed a maximum at 5.6–5.4 eV and a shoulder at ~4.8 eV [79], in good agreement with the absorption bands of F⁺ centers at 5.4 and 4.8 eV; as noted earlier, however, direct photoionization of these centers in Al₂O₃ in this spectral region appeared unlikely.

Although optical properties of F-type centers in wide bandgap metal oxides have been studied sufficiently both theoretically and experimentally, interpretation of the data on the spectral response subsequent to irradiation, even for simple photoreactions such as photoadsorption of O₂ and H₂, has encountered some problems.

Indeed, the spectral dependencies of the quantum yield of O₂ and H₂ photoadsorption in the visible and near-bandgap spectral regions for powdered TiO₂ reported by Cherkashin’s group [80,81], by Emeline and coworkers [77,82], and by Komaguchi et al. [83] have been digitized; results are illustrated in Figure 12 and Figure 13. The majority of the action spectra appear as a sum of two bands, the main band occurring at 3.0–2.9 eV with a half-bandwidth of 0.15–0.3 eV, independent of the type of gas used (O₂, H₂ or CH₄) and irrespective of the crystalline structure of TiO₂ (rutile or anatase). Spectra 2 and 3 in Figure 12 and 3 and 4 in Figure 13 display an additional band (seen as a shoulder) with maximum at ca. 2.5–2.7 eV. The action spectra of the photogeneration of paramagnetic O₂^−• (Figure 12, Curve 4) and [O^–O₂] species (Figure 13, Curve 5) show broad spectra with a single maximum around 2.6–2.8 eV.

The action spectra of the photogeneration of electrons (photoadsorption of O₂ or photoformation of O₂^−•) or of the photogeneration of holes (photoadsorption of H₂ and CH₄ or the photoformation of [O⁻O₂]) originate from excitation of intrinsic defects, which have a biographical origin because all samples examined were nominally pure (undoped). Note that the rutile specimen displaying the broad action spectrum (Figure 13, Curve 5) had been strongly reduced by a H₂ heat treatment at 773 K; it showed an absorption spectrum with a well-resolved shoulder at 2.7 eV and a broad absorption in the near-IR region (after bleaching the initial absorption in the presence of O₂) [83].

Four principal reasons led Kuznetsov and Serpone [84] to model the photoactivation of (modified) TiO₂ in the visible region: (1) the spectral position of the bands constituting the action spectra (i.e., bands at 3.0–2.9 and 2.5–2.7 eV); (2) the coincidence of the bands for the photo- generation of electrons and holes with the latter allowing the authors to propose that photoexcitation of intrinsic defects leads to the simultaneous generation of electrons and holes (electron–hole pairs); (3) the spectral position of the absorption bands at 2.95–2.75 eV and at 2.55–2.50 eV in the spectra of reduced TiO₂ attributed to F centers [84]; and (4) the known regularities of photoexcitation of F type centers and consequent physical processes. In this regard, the authors [84] proposed a simple reasonable mechanism to interpret the experimental results reported in Figure 12 and Figure 13. In accord with their study of TiO₂ specimens, the first photophysical event of photoinduced absorption and photoreactions was light absorption by the F center that led to the formation of the corresponding excited state F* (Equation (52)).

F + hν_AB → F*

(52)

F* → F⁺ + e⁻_cb

(53a)

F* + Δ → F⁺ + e⁻_cb

(53b)

F⁺ + (O²⁻)_vb → F + (O⁻)_vb

(54)

F* + (O²⁻)_vb →F + (O⁻)_vb + e⁻_cb

(55)

If the ground state of the F center lay within the bandgap near the VB band, then F* should lie within or close to the CB band. The energy difference (E_bg – hν_AB) was only about 0.2–0.3 eV between the E_bg of the rutile/anatase TiO₂ and the AB1 band, and 0.55–0.60 eV between the E_bg and the AB2 band. The excited F* center transforms spontaneously into the F⁺ center and a conduction band electron (Equation (53a)), or else it does so through thermal stimulation (Equation (53b)). Formation of a hole supposes the capture of an electron by the F⁺ center from the VB band, i.e., from the O²⁻ ion (Equation (54)). Equation (55) describes the synchronous formation of an electron–hole pair. Hence, Equations (52)–(55) infer the photogeneration of both CB band electrons and VB band holes under visible light irradiation, and recovery of the centers of light absorption during irradiation.

3.1.2. Defects Related to Cation Vacancies (V_m)

The main type of defects in the cation sub-lattice of metal oxides are defects related to cation vacancies, V_m, which have one or two holes trapped, i.e., V⁻ or V° centers, respectively. The atomic and electronic structure, the mechanism of optical excitation, and thus optical properties of hole V-type centers are defined by hole localization (trapping) at one of several equivalent oxygen ions surrounding the metal vacancy. The term bound small polaron also applies to emphasize the interconnection of carrier (hole) trapping at a single site and the stabilization of the hole at this site by a lattice distortion. INDO calculations [61] showed that, in the V⁻ center of MgO, the O⁻ ion is closer to the cation vacancy by 3% of the Mg–O distance in the perfect crystal. In the V° center, the two O⁻ ions that can capture two holes are located on the opposite side of the cation vacancy; they are shifted inward toward the vacancy by 2% of the regular Mg–O distance.

It is clear from the brief description of optical properties of photocatalytically active metal oxides that, in addition to the photoexcitation of solids in the fundamental absorption bands, there are many other pathways for the generation of electrons and holes. Concurrently, the same defects can play either negative or positive roles with respect to heterogeneous photocatalysis. For instance, they can act as recombination centers that reduce the concentration of photocarriers. However, being localized at the surface of photocatalyst particles, the defects are more likely to play the role of surface-active centers in photocatalytic processes.

3.1.3. Photogeneration, Recombination and Trapping of Charge Carriers in Photoactive Solids

Photoexcitation of solid photocatalysts in their fundamental absorption bands is the most appropriate type of photoexcitation in heterogeneous photocatalysis, as both electrons (reductants) and holes (oxidants) are generated. Concurrently, the generation of carriers also occurs as a result of light irradiation at wavelengths corresponding to extrinsic absorption bands, i.e., in the longer wavelength spectral region with respect to the fundamental absorption edge of the solids. The same is true of dopant-sensitized photocatalysts.

Recombination of electrons and holes can occur by two pathways: (i) direct recombination of free electrons from the conduction band with free holes from the valence band; and (ii) indirectly through the participation of defects (recombination centers). In the latter case, sequential trapping of free carriers of opposite charge by recombination centers takes place in each recombination cycle. The quasi-stationary concentrations of carriers established under stationary illumination of the photocatalyst depends on the concentration and trapping cross-section of the recombination centers. When photogeneration of carriers takes place in the bulk of the photocatalyst particle (a typical case), photoelectrons and photoholes reach the surface via diffusion. On the other hand, a fraction of the carriers is captured by various carrier traps, both in the bulk and at the surface of the solid particle in sufficiently high number, provided that trapping cross sections of one carrier takes place relatively slow (so-called deep traps of the color center type). Deep traps that capture the carriers may be centers of the V-type or of the F-type, depending on the charge of the captured carrier. Surface traps of this kind play the role of photocatalytic active centers. In relation to the chemical interaction with molecules, it is useful to distinguish two states of such centers—the inactive state (without trapped carriers) and the active state (trap is occupied by either an electron or a hole).

3.1.4. Trapping of Carriers by Defects

Carrier trapping by a defect is a good example of a perfect inelastic collision of carriers with the defect. The effectiveness of carrier trapping in solids is characterized by the phenomenological trapping cross-section, σ_tr (Equation (56)).

$$\frac{dN}{dt}={\sigma }_{tr}vn$$

(56)

where dN/dt is the number of carriers trapped by the defect per unit time; v is the mean velocity of the carriers (v ≈ 10⁷ cm s⁻¹); and n is the concentration of thermal carriers (or stationary concentration of photocarriers). The quantity dN/dt has dimensions and notion of frequency; its inverse can be treated as the lifetime of a defect (unoccupied by a carrier) relative to a trapped carrier.

Trapping cross-sections are determined by the properties of the defects; they vary within a rather wide range from 10⁻²¹–10⁻²⁰ cm² (trapping is energetically favorable but an activation barrier exists) p to 10⁻¹¹–10⁻¹⁰ cm² (strongly attractive Coulomb center; scattering of excess energy is effective). Typical values of σ_tr lie in the range between 10⁻¹⁶ and 10⁻¹⁵ cm² for neutral defects.

3.1.5. Stationary Concentration of Photocarriers and Band-to-Band Recombination

Figure 14 outlines the generation and recombination of carriers of various kinds that determine the stationary concentration of photocarriers in wide bandgap solids [85].

In general, both electrons and holes generated in wide bandgap solids on absorption of photons with hν > E_bg initially transit into states at some energy distance from the bottom of the CB and the top of the VB, respectively (Step 1 in Figure 14a). Thermal equilibrium between the crystal lattice and the photocarriers is established as a result of relaxation of the energy and momentum of carriers within the timescale of ~10⁻¹⁰ s (Steps 2 and 2′). Subsequent lowering of the energy of carriers occurs via recombination and trapping. For ideal (non-defective) solids, only radiative and nonradiative band-to-band recombination is possible (Steps 3 and 3′). The nonradiative electron transitions CB → VB in wide bandgap solids (E_bg ≥ 3 eV) have low probability since participation of many phonons with energy of 0.1 eV is required in the process. In general, the higher is E_bg, the higher is the probability of radiative transitions in solids with emission of photons manifested as an inter-band or edge luminescence. At the same time, the probability of radiative band-to-band transitions is rather low, since the momentum conservation law (see above) requires that the condition similar to that for band-to-band direct transitions be satisfied. Typically, band-to-band luminescence in wide bandgap solids with photon energy close to E_bg (Step 3, Figure 14a) is detected at moderate intensities of the exciting light for rather perfect crystals, for which competitive radiative recombination of carriers through defects (recombination centers, R) is suppressed, or in the case of intense photoexcitation of the crystal when a high concentration of photocarriers is achieved in the solid.

A more detailed treatment of band-to-band recombination was outside the scope of this review article, but suffices to note that band-to-band recombination in line with band-to-band optical excitation in TiO₂-based heterogeneous photocatalysis is often treated following Equations (57) and (58).

TiO₂ + hν → (e⁻ + h⁺)

(57)

(e⁻ + h⁺) → TiO₂ (band-to-band recombination)

(58)

With regard to the effectiveness of band-to-band recombination, one should use caution in believing the bimolecular stage described by Equation (58) as being real (not symbolic) in kinetic studies of heterogeneous photocatalytic reactions, even though it can easily explain experimental non-linear dependencies of reaction rates on light intensities, which typically follow a square-root dependence.

3.1.6. Recombination of Carriers via Defects

Recombination of photocarriers via recombination centers is the main pathway of carrier recombination in imperfect wide bandgap solids. In this case, a given recombination center R subsequently captures a free electron and a free hole, or vice versa, in a single recombination cycle (Steps 4 and 4′ in Figure 14a). In an alternative approach (without the concept of holes), recombination via defects means a consequent two-step transition of electrons from the CB to an empty state of the defect and from the defect to an empty state of the VB, or from the defect to the VB and then from the CB to an empty state of the recombination center R. The excess energy of the electron is dissipated at both steps of the recombination. Typically, the dissipation of excess energy at one of the two stages occurs via a nonradiative pathway with the assistance of existing phonons. Normally, the probability of nonradiative transitions increases as the number of emitted phonons decreases so that the energy levels corresponding to efficient recombination centers typically lie near the middle of the energy gap in wide bandgap solids. Radiative transitions at any stage of the two-step recombination via a center R are responsible for recombination luminescence in solids.

3.1.7. Trapping of Carriers with Formation of Centers Similar to Color Centers

In addition to the recombination centers R, other types of centers can be distinguished in wide bandgap solids—e.g., the color centers of the V- and F-type and shallow (or thermal) traps (Figure 14b). Discrimination between centers is rather arbitrary. Following the work of Siline and Trukhin [86], we may assume that a center captures photocarriers with a probability p and loses them with probability q (units: number of events per unit time). The rate of formation of defects in a state with a trapped carrier dn/dt (units: cm⁻³ s⁻¹) is then given by Equation (59) with N₀ defining the initial concentration of empty trap centers.

$$\frac{dn}{dt}=p\left(N_0-n\right)-qn$$

(59)

Equation (59) leads to the kinetics of formation of filled traps n_(t) given by Equation (60), and to the stationary concentration of filled traps n_∞ by Equation (61),

$$n_{(t)}=N_0\frac{p}{p+q}(1-e^{-(p+q)t})$$

(60)

$$n_{\infty }=N_0\frac{p}{p+q}$$

(61)

Three possible events can occur for a carrier trapped at a trap center (Figure 14c):

• Stage I, recombination via trapping of carriers of the opposite sign;
• Stage II, thermal emission of carriers into the corresponding band; and
• Stage III, photo-ionization of the trapped carrier by absorption of photons.

Accordingly, q in Equation (61) is given by,

q = q_r + q_th + q_ph

(62)

For simplicity, if the process defined by Stage III were neglected (i.e., q_ph = 0), and if q = (q_r + q_th) >> p, and q_r >> q_th, then n_∞ << N₀. In other words, filled traps do not accumulate in a solid in numbers compared with the initial number of empty traps N₀. Typically, these deep trap centers with high cross-section of carrier trapping that do not accumulate in illuminated solids are the recombination centers. In contrast to deep centers, carriers trapped in shallow traps with energy ~kT from the CB (electrons) or from the VB (holes) can be detrapped via thermal emission with probability q_th such that q_th ~ exp(–E/κT) >> q_r.

Deep centers with a low probability of trapping a second carrier (p ≈ q) accumulate in illuminated solids in sufficiently high number (n_∞ ≈ N_o). Historically, they have been referred to as color centers. The ultimate concentration of color centers can reach values of n_∞ = 10¹⁷–10¹⁸ cm⁻³ for some samples of wide bandgap solids. The term color center originated from the accumulation of such traps in crystals that display absorption bands in the visible spectral region. The result of these traps is that transparent (white when powder) solids become colored under UV illumination, with the color tending to be rather stable at ambient temperature. Such solids can be discolored by calcination, or by thermal bleaching (probability of this process given by q_th ≅ exp(−E/κT) is high), or partly by photobleaching (q_ph is also high) via illumination of the crystalline specimen with light at the wavelengths corresponding to the absorption bands of the color centers.

3.1.8. Lifetime and Concentration of Free Charge Carriers

A major factor affecting processes in heterogeneous photocatalysis is the stationary (surface) concentration of charge carriers. As such, a consideration of some basic approaches that describe the processes responsible for the charge carrier concentration is worth noting.

The spatially uniform photogeneration of charge carriers occurs at a constant rate of generation, g, in some space of the solid bulk. The rate of the temporary alteration of the charge carrier concentration is then given by Equation (63),

$$\frac{dn}{dt}=g-\frac{n}{\tau }$$

(63)

where τ is the lifetime of the free charge carrier, independent of the charge carrier concentration. In the case of several types of trapping and recombination centers, the lifetime of the charge carriers can then be expressed by Equation (64), provided that the various centers do not interact with each other. Otherwise, the determination of the lifetime becomes rather more complex [87].

$$\frac{1}{\tau }={\displaystyle \sum _i}\frac{1}{{\tau }_i}$$

(64)

where τ_i = 1/σ_tr vN_i is the lifetime of the charge carrier with respect to trapping by the defects of a given i-sort; σ_tr is the trapping cross-section; v is the velocity of the charge carrier; and N_i is the concentration of the defects of a given sort.

After an initial period of irradiation, recombination centers and trapping centers are filled with charge carriers to a stationary level; the lifetime of the charge carriers becomes pseudo-constant and its value determines the stationary concentration of the free charge carriers. Thus, dn/dt = 0, and the stationary concentration of charge carriers is then (Equation (65)),

$$n_{\infty }=g\tau$$

(65)

Obviously, the higher the concentration of recombination centers and the larger the trapping cross-section are, the shorter is the lifetime and the smaller is the concentration of free charge carriers. For semiconductors and isolators, the lifetime of photoinduced charge carriers can vary in a wide range from picoseconds (10⁻¹² s) to milliseconds (10⁻³ s).

When stationary conditions are established for the charge carriers, one fraction of the carriers remains in the delocalized state, while the other fraction is trapped by the defects. The relation between charge carriers obeys the electroneutrality principle (Equation (66)),

$$n_e+n_{tr}^e=n_h+n_{tr}^h$$

(66)

where n^e_tr and n^h_tr are the concentrations of electrons and holes, respectively, trapped by the defects.

The lifetime of a trapped charge carrier is determined by the efficiency of either thermo- or photo-ionization of the center or by the efficiency of the recombination event. When thermo-ionization dominates, the lifetime τ_th (Equation (67)) varies from picoseconds for shallow traps with depth comparable to kT, to infinite time for color centers in wide bandgap insulators.

$${\tau }_{th}=\frac{1}{q_{th}}\approx {10}^{-12}{e}^{(\frac{E}{kT})}$$

(67)

However, when the lifetime of the trapped charge carrier is determined by recombination (Equation (68)) then

$${\tau }_r=\frac{1}{q_r}=\frac{1}{{\sigma }_{r}vn}$$

(68)

where σ_r is the recombination cross-section, v is the velocity of the charge carrier of the opposite sign, and n_i is the concentration of the charge carrier of the opposite sign. The higher is the concentration of the opposite charge carrier, the shorter is the lifetime of the trapped charge carrier.

Charge carrier trapping and recombination processes determine the stationary concentration of charge carriers at the surface. Moreover, considering that surface defects with trapped charge carriers can act as surface-active centers that initiate surface chemical sequences, the lifetime of trapped charge carriers on such surface defects corresponds to the lifetime of the chemically active states of the surface-active centers. Recombination processes then return the surface-active centers to their (initial) chemically inactive ground states.

4. Applied Photocatalysis: Laboratory-Scale deNOxing of NO_x Agents (NO & NO₂)

As discussed earlier, nitrogen oxides (NO_x) are major atmospheric pollutants that play an important role in atmospheric chemistry, and have been the object of a significant number of investigations toward their minimization, if not complete removal from the environment. The concentration of NO_x in polluted urban air is around 100 ppbv, whereas in the unpolluted troposphere, it ranges from 10 to 500 pptv [88]. Recall that NO_x are emitted primarily from artificial sources (e.g., traffic, coal burning boilers, thermal power plants, and industries of various sorts) and from natural sources (e.g., biological degradation in soil and from lighting thunder). NO_x participate in various environmental processes: for instance, in the formation of acid rain; in the greenhouse effect in synergy with sulfur oxides; in the formation of photochemical smog in the presence of CO and VOCs; in the depletion of stratospheric ozone; and in the formation of peroxyacetyl nitrates (PAN), all of which have negative effects on ecosystems and lead to non-insignificant human health issues. With regard to the latter, NO_x pollutants cause problems in the respiratory tract that include lung edema and the reduction of the oxygen-caring capacity of blood—e.g., in the transformation of hemoglobin to methemoglobin.

No wonder then that significant efforts have been expended to reduce environmental NO_x agents back to N₂ via a thermal technology using a variety of reductants (e.g., CO, hydrocarbons, H₂ and NH₃) in what is known as Selective Catalytic Reduction (SCR). While reduction of NO occurs around 100 °C in the presence of H₂ and a Pd-supported catalyst [89], other reactions require significantly greater temperatures. In fact, reduction of NO to N₂ through selective catalytic reduction with NH₃, and thus potentially treat NO_x agents, the costs of the SCR technology for the construction and operation of a facility to treat NO_x pollutants, together with the required consumption of energy, may prove prohibitively high. Nonetheless, despite the many efforts to eliminate the NO_x emitted from the various sources noted earlier by SCR, the fact remains that the concentration of NO in air in Japan was nearly constant throughout the 1980s, and was often higher than the air quality standard set for NO₂, principally along heavily trafficked roads in densely populated areas [90]. This led to the development of a new technology for the disposal of NO_x at sub-ppm level from air and from trafficked roads and tunnels, and other environmental sources that emit NO_x.

Recognition that plants and micro-organisms can easily consume nitrite (NO₂⁻) and nitrate (NO₃⁻) ions as raw materials for nitrogen assimilation provided a further impetus to examine alternative technologies to achieve a practical removal of dilute NO_x agents from the environment using sunlight (UV-Visible) radiation at significantly lower costs. In this regard, Takeuchi and Isubuki [91] investigated the dry deposition of NO_x onto the ground and found that the rate of adsorption of NO_x on some soil particles was enhanced by photoillumination. Of the metal oxides constituting the soil particles, TiO₂ showed the highest activity for NO_x adsorption under photoillumination with ca. 60% of NO_x being captured as nitric acid (HNO₃) on the surface of TiO₂ particulates. Accordingly, the authors thought that the photocatalytic oxidation of NO_x to HNO₃ by illuminated TiO₂ might be most advantageous to treat dilute environmental NO_x, as any extra reactants such as NH₃ were not required and HNO₃ could be trapped on the surface as nitrates.

One of the first studies to examine the fate of one of the NO_x agents, namely NO, in the presence of (Degussa) P-25 TiO₂ exposed to UV irradiation was reported in 1984 by Courbon and Pichat [92] who exposed isotopically labeled N¹⁸O at 295 K in the dark to pre-oxidized and pre-reduced TiO₂ powder; subsequent to UV illumination resulted in three phenomena: photoadsoprtion, photoexchange, and photodecomposition of NO to yield N₂O and, to a lesser extent, N₂. The formation of N₂O + N₂ corresponded to a photodecomposition of ca. 15% of the NO pressure (decrease) for a pre-reduced titania sample and ca. 20% for a pre-oxidized titania; N₂ formed only at the beginning and the percent N₂¹⁶O produced was initially greater for the pre-oxidized titania sample. This early study [92] confirmed that illumination of TiO₂ with UV light considerably increased the ease of detachment of surface oxygen atoms, as the isotopic hetero-exchange of N¹⁸O occurred at room temperature, while it required higher temperatures in the absence of bandgap (3.2 eV) illumination of the mostly anatase TiO₂. Adsorbed oxygen species were involved, as pre-oxidized titania exhibited higher initial efficiency; however, the instantaneous exchange with a pre-reduced titania sample in H₂ at 723 K showed that detachment also involved surface oxygen atoms that were replenished from NO. Another aspect of this study was the corroboration of the direct involvement of O⁻ species in photocatalytic oxidations over TiO₂ and other n-type semiconductors, since NO and O₂ played similar roles in yielding dissociated oxygen species active in both oxidation and oxygen isotopic exchange.

A later study (1985) by Hori and coworkers [93] demonstrated that NO₂⁻ ions are oxidized to NO₃⁻ with or without O₂ in aqueous suspensions of some semiconductor powders (Ag₂O, PbO, anatase TiO₂, Si, ZnO, SnO₂, CdS, and Bi₂O₃) under bandgap illumination; with TiO₂, 96% of nitrite was oxidized to nitrate in the presence of oxygen. Along similar lines, Anpo and coworkers [94] found that Cu⁺ ions on SiO₂ (Cu⁺/SiO₂ catalyst) could decompose NO molecules photocatalytically and stoichiometrically into N₂ and O₂ at 275 K, which they attributed to the significant role played by the excited state of the Cu⁺ species; the photoreaction involved an electron transfer from the excited state of the Cu⁺ ion into an anti-bonding π orbital of the NO molecule within the lifetime of its excited state. The relationship between the local structures of Cu⁺ ions in zeolite and their photocatalytic reactivity in the decomposition of NO_x into N₂ and O₂ at 275 K was reviewed by Anpo and coworkers [95] after which Anpo’s group [96] reported on the metal ion-implantation of TiO₂ with various transition metal ions that subsequent to calcination in oxygen at ca. 723 K resulted in a large shift of the absorption edge of TiO₂ toward visible light regions depending on the amount and type of metal ions implanted; the resulting metal ion-implanted TiO₂s proved active in the photocatalytic decomposition of NO to N₂, O₂ and N₂O at 275 K under irradiation with visible light at wavelengths longer than 450 nm.

Following their 1989 report [91], Ibusuki and Takeuchi [97] examined the photocatalytic destructive oxidation of NO to NO₃⁻ using a mixture (200–250 mg) of TiO₂, activated carbon (AC) and Fe₂O₃ particles located in a flow-type photochemical reactor system (Figure 15) that was photo-illuminated by a cylindrical bank of 12 black lights (wavelength: 300–400 nm) [97]. The AC and Fe₂O₃ had a remarkable effect in increasing the catalytic activity for NO_x removal, likely due to their high adsorptive activity for NO and NO₂. The authors inferred that photo-illuminated TiO₂ generated reactive oxygen species that oxidized NO and NO₂, respectively, to NO₂ and NO₃⁻, while activated carbon trapped NO₂ to allow enough time for TiO₂ to oxidize NO₂ to NO₃⁻; it appears that Fe₂O₃ acted as a promoter for more NO/NO₂ molecules to be adsorbed on the surface of the titania photocatalyst [97].

In a further study, the Takeuchi group [98] examined the use of TiO₂ to eliminate NO_x in open air with the photocatalyst being activated by sunlight, but noted, however, that in so doing desorption of NO₂ occurred during the oxidative removal of NO; the NO₂ also needed to be suppressed as it is also a regulated pollutant. Although NO_x adsorb on activated carbon to be oxidized ultimately to NO₃⁻, development of an activated carbonaceous photocatalytic material proved difficult. Accordingly, recognizing that thin films have many micropores they designed and prepared TiO₂ thin film photocatalysts by a dip-coating process using titanium alkoxide as the TiO₂ precursor and the polymer additive polyethylene glycol (PEG) of different molecular masses (PEG-300, PEG-600, PEG-1000) to give TiO₂-PX films with thicknesses of 1.0, 0.5 and 0.25 μm after calcination of the films at 450 °C for 1 h deposited on silica-coated glass plates.

Table 1. Extent of NO removal over 1.0 μm thick TiO₂-PX thin films irradiated at 365 nm (illuminated area, 100 cm²; light irradiance, 0.38 mW cm⁻²) for 12 h in a flow-type reactor (flow rate, 1.5 L min⁻¹; initial concentration of NO, 1.0 ppm; dry air).

Photocatalytic Thin Film	BET Surface Area (m2 g−1)	Average NO Removal (%)
TiO2-P0	112	71
TiO2-P300	104	65
TiO2-P600	118	70
TiO2-P1000	141	81

summarizes the extent of NO removal [98]. Adsorbed NO was photooxidized to NO₂ by the thin films, while the produced NO₂ was re-photooxidized to NO₃⁻ before it desorbed from the film surface.

Following reports that TiO₂ prepared by high-temperature hydrolysis of titanium tetra-alkoxides, Ti(OR)₄, in a hydrocarbon solvent was very active toward the photocatalytic dehydrogenation of iso-propanol in aqueous media under deaerated conditions [99] and mineralization of acetic acid under aerated conditions [100], Hashimoto et al. [101] prepared TiO₂ by the hydrolysis of titanium alkoxide in a hydrocarbon solvent, followed by calcination at various temperatures; the titania calcined at 300 °C proved most active for the photocatalytic oxidation of NO (

Table 2. Crystal size and surface area of the titania photocatalysts together with the rate of formation and the quantities of the O₂^−• radical anions **.

Photocatalyst	Calcination T (°C)	Crystal Type	Crystal Size (nm)	BET Surface Area (m2 g−1)	Rate of O2−• Formation (μmol min−1)	[O2−•] (μmol g−1)
TiO2	300	Anatase	10	133	2.9	7.2
TiO2	550	Anatase	18	78	0.80	3.7
TiO2	700	Anatase	26	34	0.06	0.25
TiO2	800	Anatase + Rutile	47	8	0.01	0.07

** Adapted with permission from Hashimoto et al. [101]. Copyright 2000 by Elsevier Science S.A. (License No.: 4453260419845).

), in comparison with P-25 titania. The photocatalytic oxidation was carried out in a fixed bed continuous flow Pyrex-glass reactor under atmospheric pressure with the TiO₂ (0.12 g) UV-irradiated with a 10-Wh black light; air contained 10 ppm of NO; flow rate, 110 mL min⁻¹. IR spectral results indicated that UV irradiation promoted the oxidation of NO in the presence of oxygen to yield nitrate species, while the data from ESR measurements for oxygen radicals showed that UV irradiation increased the number of O₂^−• adsorbed on the surface of titania in the presence of oxygen. These O₂^−• species vanished simultaneously with their exposure to NO, whereas the spectral intensity of the radical generated from secondary products of O⁻ showed no change. The number of O₂^−• radical anions generated by UV irradiation reflected the photocatalytic oxidative activity of titania toward the oxidation of NO. The rate of formation of O₂^−• and the number of free electrons induced by UV irradiation decreased significantly with an increase in post-calcination temperature (Table 2) [101].

A photocatalytic reaction that takes place in a gas/solid reactor necessitates both the exposure of the catalysts to light irradiation and good contact between reactants and catalyst. In this regard, Lim and coworkers [102] noted that a two-dimensional fluidized-bed photoreactor not only increased the contact of catalyst and gas, but also enhanced UV light penetration compared with a packed bed reactor in which light could not easily penetrate the interior of the catalyst bed, so that it was important to design and fabricate a fluidized-bed photoreactor with higher light throughputs and lower pressure drops. Accordingly, they used: (i) an annular flow-type photoreactor; and (ii) a modified two-dimensional fluidized-bed photoreactor to examine the photocatalytic decomposition of NO. In the first case, two serial annular flow photoreactors were used to increase contact time between the gas and the photocatalyst (Figure 16).

By comparison, the modified two-dimensional fluidized-bed reactor (Figure 17) consisted of an annular-type reactor made of a larger quartz glass tube (internal diameter, 30 mm; height, 400 mm) in which a small diameter quartz tube (inner diameter, 20 mm; height, 375 mm) was located at the center of the larger tube such that the thickness of the annulus in the bed was 5 mm [102]. A quartz filter (100-mesh size) was used to distribute a uniform fluidization of the catalyst; a square mirror box surrounded the photoreactor to minimize loss of light irradiation and to improve utilization of reflected and deflected light.

In their study [102], the authors examined the effects of gas-residence time, initial NO concentration, reaction temperature and UV light source on the photocatalytic decomposition of NO carried out in the annular flow-type reactor. P-25 titania powder was used to cover a quartz tube (430 mm) by dipping it into a stirred 5% TiO₂ slurry solution and then air-dried for 24 h, after which the TiO₂-coated quartz tube was fired in a high-temperature furnace at 400 °C for 1 h; TiO₂ coating was repeated several times until the amount of TiO₂ deposited on the quartz tube reached 0.10 g. The quartz tube had been sandblasted previously to create a granular texture to anchor the fine TiO₂ powder. In addition, precursor solutions for coating TiO₂ on silica gel were prepared using titanium ethoxide as a precursor to prepare the TiO₂ sample.

A gas stream (200 mL min⁻¹) of 138 ppmv NO in He in the annular flow-type reactor was irradiated by four UV lamps without TiO₂ photocatalyst at ambient temperature for 140 min with no variation in the NO concentration; in the presence of TiO₂, however, irradiation led to the decomposition of NO with formation of NO₂, N₂O and N₂ products [102]. The reaction rate followed the power law R = R_o Iⁿ with n = 0.48 and n = 0.87 depending on the UV intensity of the germicidal white lamp (254 nm) and the fluorescent black lamp (365 nm), respectively (gas flow rate, 100 mL min⁻¹; TiO₂ loading, 0.1 g; reaction temperature, 311 K; initial NO concentration, 50 ppm). Adsorption of nitrate on the surface of the photocatalyst increased with irradiation time leading to the deactivation of the photocatalyst. The decomposition of NO decreased linearly on increasing the initial NO concentration and on decreasing the residence time of gas in the photoreactor, so that it was necessary to increase the residence time of the gaseous reactant to provide effective contact of UV light, gaseous reactant and photocatalyst to obtain higher NO decomposition in the annular photoreactor.

In the modified two-dimensional fluidized-bed photoreactor, four reaction conditions (without TiO₂/SiO₂ and UV lamp on/off, with TiO₂/SiO₂ and UV lamp-on/off) were tested to confirm whether the decomposition of NO really took place by a photocatalytic process. Indeed, in the presence of TiO₂/SiO₂ and UV lamp-on, the NO concentration decreased indicating that it was in fact decomposed [101]. Decomposition of NO increased with decreasing initial NO concentration and increasing gas-residence time; the reaction rate increased with increasing UV light intensity. Clearly, the modified photoreactor displayed efficient contact between photocatalyst and reactant gas with good transmission of UV-light and, consequently, increased the NO decomposition efficiency (>70%) compared with the annular flow-type photoreactor. Hence, the former photoreactor was an effective tool with which to carry out significant NO decomposition with efficient utilization of photon energy [102].

Anpo and coworkers [96] had earlier prepared a TiO₂ photocatalyst that subsequent to the implantation of Cr ion and upon irradiation with visible light (>450 nm) decomposed NO into N₂, O₂, and N₂O under O₂-free conditions. Additionally, the Cr ion-implanted TiO₂ catalyst displayed the exact same photocatalytic efficiency as the original TiO₂ catalyst, albeit under UV irradiation. As a follow-up to this study, Nakamura et al. [103] examined the role of oxygen vacancies in the removal of NO under an oxidative atmosphere using a commercial TiO₂ (Ishihara ST-01; 100% anatase; crystallite size, 7 nm; nominal specific surface area, 300 m² g⁻¹) and hydrogen plasma-treated TiO₂ powders; the latter was photoactive up to 600 nm without a decrease in UV light activity. Reactions to remove 1.0 ppm of NO were carried out in a Pyrex glass flow reactor (500 cm³) with irradiation from a 300-W Xe light source; the UVA (315–400 nm) irradiance at the photocatalyst surface was 0.03 mW cm⁻² (flow rate, 1500 mL min⁻¹; TiO₂ loading, 0.20 g; total pressure, 760 Torr)—no removal of NO occurred without the metal oxide photocatalyst. NO was converted mainly to NO₃⁻ (also less than 2% NO₂⁻ formed) by oxidation over the TiO₂ powder; NO₃⁻ ions accumulated on the catalyst surface. Electrons trapped in oxygen vacancies in the plasma-treated TiO₂ were detected under visible light irradiation (F-type color centers; ESR measurements displayed a signal at g = 2.004) with the number being proportional to the percent of NO_x removed, which suggested that the number of trapped electrons determined the activity of the photocatalytic oxidation of NO to NO₃⁻. The visible-light photocatalytic activity of the plasma-treated TiO₂ was due to photoexcitation of the F-type color centers with energy levels within the forbidden bandgap of the metal oxide (see Figure 18).

The Anpo group [104] investigated the photocatalytic decomposition of NO_x (NO and NO₂) on five well-characterized standard reference ultrafine powdered TiO₂ photocatalysts (grain size, 0.02–1 μm) denoted TiO₂ (JRC-TIO-2, -3, -4, and -5) supplied by the Catalysis Society of Japan (properties summarized in

Table 3. Physicochemical properties of the standard reference TiO₂ photocatalysts (JRC-TiO-2, -3, -4, and -5) supplied by the catalysis Society of Japan **.

Catalysts (JRC-TiO-)	Surface Area (m2 g−1)	Acid Concentration (μmol g−1)	Relative –OH Concentration	Bandgap (eV)
2 (anatase)	16	6	1.0	3.47
3 (anatase)	51	22	1.6	3.32
4 (anatase)	49	5	3.0	3.50
5 (rutile)	3	7	3.1	3.09

** Reproduced with permission from Zhang et al. [104]. Copyright 2001 by Academic Press (License No.: 4453251363124).

) in a large-scale continuous flow reaction system (Figure 19a) with high efficiency. Special attention was expended on the effects of pretreatment and reaction conditions on the reaction and conversion rates of NO. The authors established that surface hydroxyl groups played a significant role as active sites in the decomposition of NO.

When used on a large scale for long periods in photoreactions, photocatalysts tended to lose, albeit gradually, their photocatalytic activity. In Anpo’s study [104], after 2 h, conversion of NO for each photocatalyst leveled off and dropped to between 0.25 and 0.20 the photocatalytic activity observed initially, indicating a decline in photocatalytic activity of the TiO₂ in the decomposition of NO in the absence of O₂ and/or H₂O. Reaction products in the flow reaction system were N₂, O₂, and N₂O, just as occurred in a closed reaction system. The JRC-TIO-4 photocatalyst displayed the highest photocatalytic activity for the conversion of NO, while for the other three there were small differences: JRC-TIO-4 (11%) >>> −3 (~2%) > −5 (1.8%) > −2 (1%) (see Figure 19b). The anatase TiO₂ catalyst with the larger surface area, wider bandgap, and numerous surface –OH groups (Table 3) exhibited the highest photocatalytic reactivity in the decomposition of NO, which the authors deduced that these were the principal factors that affected photocatalytic efficiency. The increased bandgap of JRC-TiO-4 was accompanied by a shift in the conduction band edge to higher negative energies, thus moving the redox potential to more negative values thereby enhancing photocatalytic reactivity. Moreover, surface –OH groups and/or physisorbed H₂O also played a significant role in the photocatalytic reactions through the facile formation of reactive ^•OH radicals. The intensity of the incident light is also an important factor that affects the kinetics of the photocatalytic decomposition. The quantum efficiency of the photocatalytic reaction was higher at the lower intensities of the incident light, and lower at higher intensities of the incident UV light; in addition, the efficiency of conversion of NO increased with increase in the O₂ flow rate.

The activity of the JRC-TiO-4 photocatalyst was also tested by Tanaka and coworkers [105] in the photoassisted selective catalytic reduction of NO with ammonia (photo-SCR) at low temperature over irradiated TiO₂ in a flow reactor; the process was efficient and the adsorbed ammonia reacted with NO under irradiation of TiO₂ (Figure 20); note the nearly identical kinetics of formation of both N₂ and N₂O.

The total amount of N₂ formed was 0.23 mmol g_cat⁻¹, consistent with the amount of ammonia (0.24 mmol g_cat⁻¹) adsorbed over TiO₂ in equilibrium at 323 K. The kinetic experiment carried out under differential conditions in the pressure range 300 < p(NO), p(NH₃) < 2000 ppm, and the presence of excess O₂ affected the evolution rate of N₂ which depended only on partial pressure of NO; kinetics were first order on NO, and zeroth order on O₂ and NH₃, which strongly suggested that the rate-determining step was adsorption of NO to the irradiated TiO₂ adsorbing ammonia molecules [105]. To the extent that the selective catalytic reduction (SCR) with ammonia is a downhill reaction, it also proceeded in the dark at low temperature with a 20% conversion of NO. However, photoirradiation caused a remarkable enhancement of the activity: the evolution rate of N₂ gradually increased attaining a steady rate at ca. 80% conversion after 2 h of irradiation.

To achieve a further understanding of surface reactions involved in TiO₂-based photocatalysis, Dalton and coworkers [106] examined two titania samples (one of unknown source) using X-ray photoelectron spectroscopy and Raman spectroscopy to investigate the NO_x adsorbate reaction at the surface of these two TiO₂ substrates. The NO_x gas was composed of 109 ± 5 ppm of NO_x, 21.0 ± 0.4% O₂, the remaining ca. 79% being N₂; dry air was the mixer gas to dilute the NO_x (NO_x concentration, 10–100 ppm) during the reaction performed under UV exposure for 6 and 48 h in a glass vessel (ca. 3 mm thick; Figure 21) that allowed > 80% transmission of the radiation at λ = 320 nm. Formation of NO₃⁻ did not vary significantly with either exposure time or NO_x concentration. The authors [106] proposed a stepwise mechanism (Figure 22) in which the surface hydroxyls increased the efficacy of the process and participated by reacting with NO_x molecules to yield nitrate ions formed indirectly via initial reductive (formation of O₂^−• radical anions by conduction band electrons) and oxidative (formation of ^•OH radicals by valence band holes) processes.

Dalton et al. [106] concluded that TiO₂ was effective at converting NO_x agents to NO₃⁻ and that XPS proved useful in quantifying the efficiency of the reaction, while Raman spectroscopy was a quick and simple way of ascertaining the surface crystal structure of the titania. XPS confirmed only one oxidation state of Ti on the untreated TiO₂ materials; however, the O_1s peak indicated the presence of two additional components of TiO₂: Ti–OH and Ti–OH₂. After exposure to UV radiation, XPS spectra revealed nitrogen peaks attributable to organic species (also present before reaction), to some unreacted NO adsorbed on the surface, and to nitrate anions.

Reactive nitrogen (NOy) in the atmosphere consists of the sum of the two NO_x oxides (NO + NO₂) and all compounds produced by atmospheric oxidation of NO_x that include the minor species: HNO₃, HNO₂, the nitrate radical NO₃^•, N₂O₅, peroxynitric acid HNO₄, peroxyacetyl nitrate (PAN) (CH₃C(O)OONO₂) and its homologs, and peroxyalkyl nitrates (RC(O)OONO₂) [107]. Such compounds can be regarded as reservoirs of NO₂ but apparently play no critical role in the formation of ozone O₃ that the precursors NO₂ and NO do. The oxidative removal of NO over irradiated TiO₂ catalyst was examined by Devahasdin and coworkers [107] at source levels (5–60 ppm) in a thin-film photoreactor systems (see Figure 23); the process involved a series of oxidation steps through the action of photoformed ^•OH radicals (NO → HNO₂ → NO₂ → HNO₃). Light intensity increased the capability to oxidize NO (from 0 to 0.8 mW cm⁻²); the selectivity for NO₂ increased with light intensity for 5 ppm inlet NO but remained constant for 40 ppm inlet NO. The steady-state conversion of NO increased with relative humidity from 0 to 50% leveling off at higher relative humidity; the ratio of NO₂⁻ to NO₃⁻ from spent catalyst liquor decreased with irradiation time until steady state was reached.

Transient behavior of TiO₂ during the first 2 h of operation with the system setup of Figure 23 (conditions: space time: 12 s; inlet concentration, 40 ppm; light source, two 8-W black lamps; relative humidity, 50%; TiO₂ loading: 1.07 mg cm⁻²) revealed that initially the conversion of NO was very high (ca. 95% after 0.5–3 min of irradiation depending on TiO₂ loading) and decreased approaching steady state after 6 h of operation; all the nitrogen was accounted for in the gas phase: NO out (26 ppm) + NO₂ out (14 ppm) = NO in (40 ppm). Conversion of NO was 35%; gas phase mass balance showed no N₂O formed in the reaction system under steady-state conditions [106]; NO₂ selectivity remained constant at 100% for 40 ppm inlet NO and increased with light intensity for 5 ppm inlet NO with a 50% relative humidity, which suggested that for 40 ppm inlet NO at steady state, all the NO should have been converted to NO₂. However, the authors [107] believed that, for the 5 ppm inlet NO, the true steady state had not yet been reached, so that increasing light intensity caused the HNO₃ to dissociate back to NO₂ and ^•OH and to promote NO₂ selectivity from 82% to 95%. The latter inference called attention for the first time to the possible reNOxification of the nitrates produced in the deNOxification of the environment (see Section 5).

Toma and workers [108] reported using a test chamber built specifically for the TiO₂ (Degussa P-25; in powder or pellet form) photocatalytic decrease of NO_x. The experimental device consisted of three parts: (a) a chamber where gaseous NO_x were prepared in situ by chemical reaction of Cu powder with a dilute solution of HNO₃; (b) an environmental chamber; and (c) a NO_x analyzer. The pollutants were subsequently injected at ambient temperature into the environmental chamber (volume, ~0.4 m³) until the concentration of NO_x reached 1–2 ppmv. A fan ensured homogenization of the gaseous pollutants in the environmental chamber. A polycarbonate photoreactor (100 mm × 100 mm × 50 mm box) equipped with a 70 × 70 mm Plexiglas window allowed light transmission from a 15-W daylight lamp (30% UVA, 4% UVB) placed inside the environmental chamber and crossed by the NO_x flow (flow rate, 0.6 L min⁻¹); NO_x concentrations were continuously monitored with a chemiluminescence NO_x analyzer.

For small TiO₂ powder quantities, conversion rates increased proportionally reaching maximal value at 0.2 g loading of TiO₂; at higher quantities of TiO₂ the decrease in NO_x remained constant and independent of TiO₂ powder amount. After 30 min of UV irradiation (surface, 54 cm²), conversion rates were about 32–35% and 15–18%, respectively, when the mass of the catalyst varied from 0.2 to 1.2 g; maximal conversion was reached at 3.7–4 mg cm⁻² of TiO₂ powder [108]. Exposing a TiO₂ pellet surface (mass, 0.4 to 1.2 g; 54 cm²) to UV radiation from one side only led to a photocatalytic conversion of ca. 28–30% of NO and 10–12% of NO_x (Figure 24); the conversion efficiency increased with the surface area of the pellet. The amount of compressed TiO₂ powder and the thickness of the pellet had little influence on the extent of NO_x decomposition. Anatase TiO₂ showed better efficiency for the photocatalytic decrease of NO_x relative to rutile TiO₂, accounting for only 10% and 5%, respectively, for NO and NO_x removal.

We noted above that the photocatalytic decomposition of NO over TiO₂ reported in some of the literature led to the formation of N₂O as the main reaction product [92,94,98] with minor N₂, NO₂ and O₂ products. Only Anpo’s group [96] reported the selectivity of NO photodecomposition over a TiO₂ photocatalyst to yield N₂O and N₂, and no other products. According to the views of Bowering and coworkers [109], use of only TiO₂ as the catalyst is not ideal for removing NO from the atmosphere as N₂O itself is also a regulated pollutant. As such, Tanaka et al. [105] reported that photoassisted selective catalytic reduction (photo-SCR) of NO over TiO₂ with NH₃ as a reductant was very selective towards N₂ formation, with relatively small amounts of N₂O. As NH₃ is also a pollutant, it would need to be eliminated from the exhaust gas, thereby causing an increase in overall costs of a system.

Car exhaust and industrial emissions are mostly controlled using selective catalytic reduction (SCR) to convert NO_x to N₂. Accordingly, Bowering et al. [109] used CO as the reductant in eliminating NO photocatalytically with Degussa P-25 TiO₂ in a continuous flow reactor (Figure 25) with the objective to convert the NO_x preferentially into N₂ gas. The authors added TiO₂ powder (ca. 0.2 g) to acidified triply deionized water (TDW; 6 mL of 0.05 M HNO₃ in 500 mL of TDW) yielding a dispersion that was stirred for 12 h and then dried at 70 °C for 48 h, after which the resulting powders were calcined for 2 h at 120, 200, 450 or 600 °C. Subsequently, 25-mL fractions of the dispersion were evaporated at 70 °C onto degreased borosilicate glass slides; an amount of TiO₂ powder (~1 mg) was deposited on the slides and then calcined following the same methodology as for the powders.

The effect of calcination temperatures on the composition and crystallite sizes of P-25 photocatalysts, and the effect of pretreatment temperature on rate of NO conversion and selectivity for N₂ formation for NO decomposition and reduction reactions are presented in

Table 4. Effect of calcination temperatures on the composition and crystallite sizes of P-25 photocatalysts, and of the pretreatment temperature on the rate of NO conversion and selectivity for N₂ formation for NO decomposition and reduction reactions **.

Calcination T (°C)	Rutile (vol. %)	Crystallite Size (nm)		BET Surface Area (m2 g−1)	Rate of NO Conversion (μmol h−1 gcat−1)		Selectivity for N2 Formation (%)
		Anatase	Rutile		Dec.	Red.	Dec.	Red.
70	23.0	28.0	66.3	51.92	1210	657	21	46
120	23.2	28.8	64.8	50.69	1107	560	25	49
200	23.0	28.2	62.7	49.87	983	467	21	48
450	23.0	28.3	59.6	49.54	550	243	26	26
600	28.0	30.5	69.6	48.24	430	240	30	25

** Adapted with permission from Bowering et al. [109]; Copyright 2005 by Elsevier B.V. (License No.: 4452670211528).

[109]. Pretreatment (calcination) temperatures caused no appreciable change in phase composition; original composition (ca. 77 vol.% anatase, 23 vol.% rutile) was maintained even after treatment at 600 °C. The photocatalytic activity for both decomposition and reduction reactions decreased with increasing pretreatment temperature, which was attributed to removal of surface hydroxyl species that acted as active sites for reaction. The only products observed in the decomposition reactions were N₂ and N₂O; the selectivity for nitrogen formation remained constant (ca. 23%) regardless of pretreatment temperature. However, the presence of CO in the reaction gas had a dramatic effect on selectivity of the reactions with N₂ selectivity as high as 65%; in addition, an increase in the CO/NO ratio led to increased selectivity for N₂ formation.

It is likely that under UV illumination electron transfer occurred from electron trapped centers into antibonding orbitals of adsorbed NO molecules, resulting in their decomposition and formation of N_(ads) and O_(ads) surface species, which can then scan the TiO₂ surface and react with other surface species (e.g., NO_(ads), N_(ads), O_(ads)) to form N₂O, NO₂, O₂ and N₂. To the extent that neither O₂ nor NO₂ was detected led the authors [109] to deduce that Reactions (69) and (70) did not occur on P-25 surfaces under decomposition conditions; the main surface reaction was Reaction (71), as N₂O was the major reaction product under these conditions.

NO_(ads) + O_(ads) → NO_2(ads

(69)

O_(ads) → O_2(ads)

(70)

NO_(ads) + N_(ads) → N₂O_(ads)

(71)

In the presence of CO on the photocatalyst surface, and under UV illumination, other reactions are possible between adsorbed CO and NO molecules together with reactions of CO with N_(ads) and O_(ads) atoms. No reaction occurred in the dark and under UV illumination without TiO₂ indicating that both TiO₂ and UV are required for adsorbed NO and CO species to react. Under decomposition conditions, the major reaction product was N₂O (~75%) with N₂ being the minor product (ca. 25%). On the other hand, under reduction conditions selectivity for N₂ formation increased (ca. 48%) at the pretreatment temperatures of 70 and 120 °C. However, at higher pretreatment temperatures, the selectivity was similar to that achieved in the absence of CO suggesting that the surface N₂ forming reaction was favored on a titanium surface rich in hydroxyl groups [109].

Germane to the previous study [109], Roy and coworkers [110] examined a photocatalytic route to destroy NO_x by developing a new Pd ion-substituted TiO₂ system (Ti_1−xPd_xO_2−δ) with which to reduce NO in the presence of CO via creation of redox adsorption sites and using anion oxygen vacancies on titania; the optimal Pd²⁺ ion concentration was 1 at.% in TiO₂ (anatase). Apparently, even though both NO and CO competed for the same Pd²⁺ adsorption sites, reduction of NO to N₂O was two orders of magnitude higher with the Ti_0.99Pd_0.01O_1.99 photocatalyst under ambient conditions than unsubstituted TiO₂; using UV irradiation with a 125-W high-pressure Hg lamp and an inlet 5000 ppm of NO in a flow-type reactor, the rate of reduction of NO was 0.53 μmol g⁻¹ s⁻¹.

The photocatalytic deNOxing activities of TiO₂, N-doped TiO₂, Fe-loaded N-doped TiO₂, and Pt-loaded N-doped TiO₂ exposed to irradiation from monochrome LED lamps at various wavelengths have been investigated in some detail by Yin and coworkers [111], unlike many studies that have typically used 100–500 Watt high-pressure Hg or otherwise Xe light sources to activate titania-based photocatalysts. Bare TiO_2−xN_x (denoted TiON) powders were prepared by treating 20-nm Ishihara ST-01 anatase TiO₂ in an NH₃ atmosphere at 600 °C for 3 h, followed by annealing at 300 °C for 2 h in humid air (N content, ca. 0.25 at.%); for comparison, the ST-01 anatase TiO₂ powder was heat-treated in air at 500 °C for 1 h (S-TiO₂)—BET specific surface areas were 57.7 m² g⁻¹ and 100.7 m² g⁻¹, respectively, for TiON and S-TiO₂ powders. The Fe- and Pt-loaded N-doped TiO₂ systems were prepared by dispersing TiON powder in a HNO₃ aqueous solution containing either Fe(NO₃)₃ or Pt(NH₃)₂(NO₃)₂ at ambient temperature, followed by stirring for 1 h, heated at 150 °C to remove the water, and then calcined at 300 and 400 °C for 2 h, respectively (loading of Fe and Pt, 0.5 wt.%; systems denoted TiONFe and TiONPt; BET areas were 61.1 m² g⁻¹ and 59.0 m² g⁻¹, respectively).

The specifics of irradiation from the four LED sources were (wavelength, irradiance): (i) red light LED (627 nm, 72.76 μmol m⁻² s⁻¹); (ii) green light LED (530 nm, 125.12 μmol m⁻² s⁻¹); (iii) blue light LED (445 nm, 76.22 μmol m⁻² s⁻¹); and (iv) UV light LED (390 nm, 73.70 μmol m⁻² s⁻¹). Different samples showed different wavelength dependencies; for instance, S-TiO₂ displayed excellent activity at 390 nm but very weak activity at 445 nm, while TiON showed excellent UV light (390 nm) and visible light-induced photocatalytic activity on exposure to 445 nm and 530 nm irradiation (Figure 26) [111]. By comparison, both TiONFe and TiONPt showed excellent deNOxing abilities even under red light irradiation at 627 nm. Specifically, TiONPt showed the highest deNO_x abilities at all light wavelength ranges: about 37.8%, 36.8%, 28.2%, and 16.0% of NO_x was removed, respectively, under continuous irradiation by monochromatic light at 390 nm (UV LED), 445 nm (blue LED), 530 nm (green LED), and 627 nm (red LED).

Because of the newly formed N_2p level within the bandgap of titania above the O_2p valence band, N-doped titania displayed an extrinsic bandgap smaller than the intrinsic bandgap of titania such that TiON absorbed visible light. In addition, TiON displayed significant chemiluminescence emission attributed to formation of singlet oxygen ¹O₂ relative to undoped S-TiO₂ which failed to display any light emission. For comparison, TiONFe and TiONPt also displayed relatively high chemiluminescence emission, albeit lower than TiON; the latter showed chemiluminescence intensity increases in the order: UV < blue < green < red.

For the undoped S-TiO₂ sample, the study of Yin et al. [111] showed a correlation between very weak chemiluminescence emission intensities and very low visible-light induced photocatalytic activity that they attributed to its relatively large intrinsic bandgap. By contrast, for TiON, TiONFe, and TiONPt, results demonstrated that the deNO_x ability decreased with an increase in chemiluminescence emission intensity. Nonetheless, the photocatalytic deNO_x activity of TiON was nearly the same as that of S-TiO₂ under 390 nm (UV LED) and 627 nm (red LED) irradiation. However, TiON exhibited greater activity than S-TiO₂ under 445 nm and 530 nm irradiation, but lower than TiONFe and TiONPt samples under every type of LED light irradiation. The authors ascribed this variation to different band structures and to the presence of Fe and Pt loaded onto the surface of the TiON.

Mechanistically, the sequence of events that led to deNOxing by these four titania samples was summarized [110] by the series of Reactions (72)–(77). Subsequent to irradiation of the titania that yields conduction band electrons (e⁻) and valence band holes (h⁺), formation of singlet oxygen ¹O₂ (Equation (76)) competes with formation of superoxide radical anions (Equation (73)) and hydroxyl radicals (Equation (74)) in air (molecular oxygen; relative humidity, ca 25%). Fe and Pt loading on TiON increased charge transfer and charge separation on the surface of the TiONFe and TiONPt photocatalysts [111].

TiO₂ + hν → e⁻ + h⁺

(72)

e⁻ + O₂ → O₂^−•

(73)

O₂^−• + H⁺ → HOO^•

(74)

H₂O + h⁺ → ^•OH + H⁺

(75)

O₂^−• + h⁺ → ¹O₂

(76)

O₂^−• + h⁺ → ³O₂

(77)

In the present context, deNOxing reportedly occurred by oxidation of the NO_x molecules via the active oxygen ^•OH and O₂^−• species (Equations (78)–(80)), whereby NO is converted to NO₂ and subsequently to NO₃⁻ ions.

NO + 2 ^•OH → NO₂ + H₂O

(78)

NO₂ + ^•OH → NO₃⁻ + H⁺

(79)

NO + O₂^−• → NO₃⁻

(80)

In a NO_x atmosphere, the NO_x molecules adsorb onto the photocatalyst’s surface and then interact with the superoxide radical anions O₂^−• to form NO₃⁻ (Equation (80)), as a result of which the NO_x molecules consume O₂^−• and delay singlet oxygen formation (Equation (76)), thereby causing the chemiluminescence emission intensity in the NO_x atmosphere to be much lower than in air. To recapitulate, Yin and coworkers [111] deduced that:

• Nanosized titania exhibited very low deNO_x ability under visible light irradiation, irrespective of their excellent UV light-induced (390 nm) deNO_x ability.
• N-doped titania displayed excellent photocatalytic activity under 445 nm and 530 nm light irradiation.
• Fe and Pt loading improved the photocatalytic activity of N-doped TiO₂ under not only UV light but also long-wavelength visible-light irradiation (λ = 530 nm and λ = 627 nm).
• Pt-loaded, N-doped titania possessed the best visible light- and UV-induced photocatalytic activity. In addition, Fe- and Pt-loaded N-doped titania exhibited relatively high quantum yields of deNOxing under long-wavelength LED light irradiation.

In their extensive 2009 review article on the catalytic abatement of NO_x in the environment, Roy and coworkers [112] focused mostly on thermal methods in the presence of suitable reducing agents, and briefly gave a short account of the alternative photocatalytic methodology at ambient conditions; summarized was also some of the earlier work reported by selected researchers noting that direct photocatalytic decomposition of NO would yield N₂ and O₂, which would indeed be the ideal outcome and sole products if that could be realized. Unfortunately, as noted above, different conditions and different titania-based photocatalysts lead to significantly different results that are worth recalling briefly. For instance,

• Anpo and coworkers showed that metal ion-implanted TiO₂ decomposed NO photo- catalytically to N₂, O₂ and N₂O at 275 K under irradiation with visible light at wavelengths longer than 450 nm [96].
• Lim et al. [102] found that the photocatalytic decomposition of NO over Degussa P-25 TiO₂ in an annular flow type reactor produces NO₂, N₂O and N₂, with the efficiency increasing with light intensity and residence time and decreasing with initial NO concentration.
• Bowering et al. [109] showed that the photocatalytic activity of Degussa P-25 TiO₂ toward deNOxing decreased with increasing pretreatment temperature.
• Roy and coworkers [110] reported that reduction of NO over the catalyst Ti_1−xPd_xO_2−δ was two orders of magnitude greater than unsubstituted TiO₂. Direct NO decomposition into N₂ and N₂O occurred via dissociation of NO in the presence of UV radiation at room temperature yielding N₂, N₂O and O₂ with the O₂ evolved reacting with NO to give NO₂ that is adsorbed by the catalyst upon formation. Prolonged NO₂ adsorption makes the surface inactive for NO dissociation; NO dissociation resumed when CO was passed to scavenge the evolved dissociated O₂ [110].

On the other hand, the seminal review article by Skalska et al. [113] presented an extensive survey of NO_x emission control technologies for three major anthropogenic emission sources: power plants, vehicles and the chemical industry, and further described new and alternative methods such as a hybrid system of SCR (selective catalytic reduction) and O₃ injection, fast SCR, and electron beam gas treatment, among others. Also described was the influence of NO_x on the environment and human health. The main focus was put on NO_x control methods applied in the combustion of fossil fuels in power stations and mobile vehicles, together with methods used in the chemical industry; the authors emphasized the implementation of ozone and other oxidizing agents in NO_x oxidation.

Following these footsteps, Heo and coworkers [114] combined photocatalysis and SCR with hydrocarbons as reducing agents (HC/SCR) to improve the activity and durability of deNO_x catalysts. The authors developed a photocatalytic HC/SCR system that exhibited high deNO_x performance (54.0−98.6% NO_x conversion) at low temperatures (150−250 °C) using dodecane as the HC reductant over a hybrid SCR system that included a photocatalytic reactor (PCR) and a dual-bed HC/SCR reactor (Figure 27). The PCR generated the highly active oxidants O₃ and NO₂ from O₂ and NO in the feed stream, followed by subsequent formation of the highly efficient reducing oxygenated hydrocarbon (OHC), NH₃, and organo-nitrogen compounds. These reductants were key in enhancing the low-temperature deNO_x performance of the dual-bed HC/SCR system containing Ag/Al₂O₃ and CuCoY in the front and rear bed of the reactor, respectively (

Table 5. Effect of catalysts on the performance of the PCR system at 150 °C. Feed gas composition: 200 ppm NO, 134 ppm C₁₂H₂₆, 6% O₂, 10% H₂O, and N₂ balance. Total flow rate = 500 mL min⁻¹; TiO₂ (anatase nanopowder) **.

Catalyst		Yield of Reductants (%)		NOx Conversion (%)
2 BaY +		Total OHCs	NH3
	TiO2	35 (84 a)	-	43
	V2O5/TiO2	60	-	29
	Au/TiO2	24	-	43
	Ag/TiO2	18	-	55
	Pt/TiO2	2	-	22
	CuCoY	19	-	37
	BaY	10	-	55
	Ag/Al2O3	16	6	54
Blank PCR (no catalyst)		45	-	-

^a Yield of OHCs in the absence of NO. ** Reproduced with permission from Heo et al. [114]. Copyright 2013 by the American Chemical Society.

). Moreover, the OHCs proved particularly effective for both NO_x reduction and NH₃ formation over the Ag/Al₂O₃ catalyst, while NH₃ and organo-nitrogen compounds were effective for the reduction of NO_x over CuCoY. The photocatalytic assisted hybrid HC/SCR system demonstrated an overall deNOxing performance comparable to that of the NH₃/SCR, thus its potential as a promising alternative to the current urea/SCR technology [114].

To the extent that the deNO_x performance of the conventional HC/SCR catalyst was enhanced by both OHCs and NH_x-containing reductants, the two representative PCR catalysts V₂O₅/TiO₂ and Ag/Al₂O₃ were chosen by the authors [114] for further examination in the PCR + HC/SCR hybrid system: the V₂O₅/TiO₂ PCR for its superior OHC formation and the Ag/Al₂O₃ PCR for its formation of NH₃ and possibly organo-nitrogen compounds as precursors of NH₃.

Table 6. Conversion of NO and yields of NO₂, N₂O, and NH₃ during the reduction of NO over the PCR + HC/SCR system. Feed gas composition: 200 ppm NO, 134 ppm C₁₂H₂₆, 6% O₂, 10% H₂O, and N₂ balance. Gas hourly space velocity of the HC/SCR monolith reactor = 16,500 h⁻¹ **.

System	Temperature (°C)	NO (%)	NO2 (%)	N2O (%)	NH3 (%)	Total NO (%)	Estimated N2 (%)
HC/SCR only	256	13	0	1	0	13	13
	303	62	0	1	1	62	60
	400	100	0	3	2	100	96
	500	53	7	1	1	46	44
Ag/Al2O3 PCR + HC/SCR	200	63	0	4	0	63	59
	250	99	0	5	4	99	91
	300	100	0	5	13	100	82
V2O5/TiO2 PCR + HC/SCR	200	42	0	4	1	42	38
	250	95	0	5	0	95	91

** Reproduced with permission from Heo et al. [114]. Copyright 2013 by the American Chemical Society.

lists the conversions of NO and NO_x, the conversion of NO_x to N₂, and the yields of NO₂, N₂O, and NH₃ during the reduction of NO with the PCR + HC/SCR system [114]. The NO_x conversion to N₂ was estimated from the conversion of NO_x and the yields of NO₂, N₂O, and NH₃ by the mass balance of nitrogen. The selectivity of the Ag/Al₂O₃ PCR + HC/SCR system for N₂ was 94%, 91%, and 82% at 200, 250, and 300 °C, respectively. The slight decrease N₂ in selectivity of the Ag/Al₂O₃ PCR + HC/SCR system at 300 °C was ascribed to increased formation of NH₃ by reaction of NO with OHCs over the HC/SCR reactor, since the PCR readily converted dodecane to OHCs. At 400 °C, the HC/SCR system alone completely reduced NO_x with high N₂ selectivity up to 96% so that the PCR could be turned off and bypassed to save energy at temperatures above 400 °C.

Key to the successful demonstration of this advanced deNO_x process was the unique design and functionality of the PCR, which led to three major conclusions [114]: (1) PCR with catalysts was very efficient for both OHC formation and reduction of NO_x because of its dual function: in situ UV-induced formation of OHC and conversion of NO_x over the catalysts; (2) blank PCR (no catalyst) was very efficient for oxidation of NO to NO₂ and HC to OHC, but was inefficient for converting NO_x because of the absence of a catalyst; and (3) Ag/Al₂O₃ PCR (with BaY + Ag/Al₂O₃) produced OHC and NH₃ as intermediates that could be used subsequently to further convert NO_x in a downstream reactor containing a dual-bed catalyst such as Ag/Al₂O₃ (for OHC/SCR) and CuCoY (for NH₃/SCR).

5. Applied Photocatalysis: Prospective Attempts at DeNOxing the Atmospheric Environment

Energy-related emissions of nitrogen oxides continue to increase worldwide, standing close to 110 Mt in 2015 with the transportation sector accounting for 52%, followed by industry (26%) and power generation (14%). China (23 Mt) and the United States (13 Mt) accounted for ca. 33% of global NO_x emissions that year. According to the International Energy Agency [115], power generation in 2015 was a major source of worldwide emissions of nitrogen oxides (14% of total NO_x) with coal being the principal fuel responsible for 70% of those NO_x emissions; burning oil to generate electricity also produced significant quantities of NO_x. Natural gas-fired plants emitted fewer air pollutants than either coal-fired or oil-fired power plants; however, in 2015, gas-fired power generation emitted close to 20% of NO_x, while biomass played a negligible role in global power generation, although, in relative terms, it performed only slightly better than coal-fired plants for NO_x emissions. Manufacturing industries and other transformation sectors (e.g., refining and mining) accounted for ca. 30% of NO_x (28 Mt) in 2015. Process-related NO_x emissions were mostly released in cement making (1.5 kt of NO_x per Mt of cement that accounted for >50% of global process-related NO_x emissions) followed by pulp and paper production (1.2 kt NO_x Mt⁻¹ of paper) (see Figure 28) [115]. Considering combustion and process emissions from a regional perspective, China was the largest emitter of NO_x, accounting for nearly 40% followed by the United States (11%). However, the United States witnessed considerable decreases in NO_x emissions in decades prior to 2015, while NO_x emissions from Chinese and Indian industries increased significantly. Together with ammonia, NO_x and SO₂ are the main precursors to formation of acid rain, which affects soil and water (with adverse impact on vegetation and animal life) and accelerates the deterioration of equipment and cultural heritage [115]. The presence of NO_x and volatile organic compounds (VOCs) in the environment leads to formation of ground-level ozone (O₃) under sunlight.

Thus, the considerable research interest witnessed over the last 2–3 decades to attenuate the extensive presence of NO_x in the environment is not surprising. In a 1999 Technical Bulletin, the United States Environmental Protection Agency (EPA) [116] described the various components that make up the NO_x pollutants, together with their properties, some of the health concerns, and how the environmental NO_x could be abated and controlled by external combustion—pollution prevention methods and add-on control technologies—that is, by non-photocatalytic technologies. In this regard, methods to reduce thermally the NO_x emissions at the origins include improved combustion techniques (e.g., fuel denitrogenation, modification to combustion methods, modification of operating conditions, and tail-end control processes) and installation of low-NO_x burners in process heaters and industrial heat and electricity generation plants. Another primary combustion technology is a fluidized-bed combustion technology for solid fuels, while end-of-pipe technologies focus on the removal of NO_x from flue gases by means of either physical separation or chemical reactions before their release to the atmosphere. Selective catalytic reduction (SCR; NH₃ as the reductant; presence of a combination of TiO₂ and V/W oxides as catalysts (Equations (81) and (82)) or selective non-catalytic reduction systems (SNR; urea or ammonia (Equations (83) and (84)) can significantly reduce NO_x in the flue gas of stationary sources into N₂ and H₂O.

Reduction of NO_x in the environment may not only involve SCR and SNR technologies, but also and of particular relevance herein is the TiO₂-based photocatalytic technology that is the subject of this section.

(a) Selective Catalytic Reduction (SCR)

4 NO + 4 NH₃ + O₂ + Cat. → 4 N₂ + 6 H₂O

(81)

2 NO₂ + 4 NH₃ + O₂ + Cat. → 3 N₂ + 6 H₂O

(82)

Conditions: Temperature: 300–400 °C; typical efficiencies: about 80%

(b) Selective Non-catalytic Reduction (SNR)

4 NO + 4 NH₃ + O₂ → 4 N₂ + 6 H₂O

(83)

4 NO + 2 (NH₂)₂CO + O₂ → 4 N₂ + 4 H₂O + 2 CO₂

(84)

Conditions: Temperature: 900–1000 °C; 40–60% reduction is obtained.

Several review articles have appeared in the last decade [117,118,119] that described, among others, some of the early attempts in the abatement of NO_x agents under indoor and outdoor experimental conditions. The 2008 article by Fujishima et al. [117] offered an overview of some highlights of TiO₂ photocatalysis, reviewed some of its origins, and indicated some useful applications: self-cleaning surfaces, water purification, air purification, self-sterilizing surfaces, anti-fogging surfaces, heat dissipation and heat transfer, anticorrosion applications, environmentally friendly surface treatment, photocatalytic lithography, photochromism of metal oxides, and microchemical systems. In the air purification application (i.e., deNOxing), the 2008 article briefly noted that some Japanese companies were considering covering roads with the TiO₂ photocatalyst, and removing the NO_x from automobile exhaust with sunlight using TiO₂-coated road bricks, prepared by mixing colloidal TiO₂ solutions with cement: the photo-road technology applied to no less than 14 different locations in Japan, one of which was the 7th belt highway in Tokyo (surface area covered, ca. 300 m²); NO_x removed from this testing area was ca. 50–60 mg per day, equivalent to NO_x discharged by 1000 automobiles (Figure 29); however, no relative efficiency was provided with respect to total NO_x in the environment.

The 2013 article by Hanus and Harris [118] entertained some innovations of nanotechnology for the construction industry, most noteworthy being improvements in concrete strength, durability and sustainability being achieved with use of metal/metal-oxide nanoparticles and engineered nanoparticles (carbon nanotubes and carbon nanofibers), as well as environment-responsive anticorrosion coatings formed using nano-encapsulation techniques.

For their part, Fresno and coworkers [119] described achievements, near-future trends and critically assessed many photocatalytic materials on the basis of knowledge accumulated in pre-2014 years as to which materials or multicomponent systems, among the multitude of developments, could be taken as a ready consolidated technology or else as more likely to become a real alternative in the short term. Germane to this, they noted that a photocatalyst could be incorporated during material manufacturing either as an additive, most often the case of construction materials, or as a coating on an already conformed cementitious surface with the mechanical resistance of the coating, the optimal amount of photocatalyst (TiO₂) and binder, their impact on the properties of the materials and their long-term performance and aesthetic durability being factors to consider in their applications as carbonation can lead to deactivation after several months of use, a point we shall emphasize later with regard to a most celebrated example: the Dives in Misericordia church in Rome (see below) [120] built with Italcementi’s TX-Active^® photocatalytic concrete [121] for self-cleaning purposes and for reduction of NO_x pollution (among others). Incorporation of anatase titania into the wearing layer, and the use of a double-layered concrete with addition of the photocatalyst to the top layer were two of the possible strategies noted for this application to reduce NO_x. Most importantly, the authors [119] emphasized the need for further assessments of the durability of the photoactive coatings, and their capability of abating other air pollutants (e.g., the VOCs).

Along these lines, the discussion below will first emphasize the results from TiO₂-based photocatalytic deNOxing the environment with the photocatalyst deposited on cementitious substrates in an indoor laboratory setting using small photoreactors set in flow-through systems and UV light sources. This is then followed by deNOxing results from various outdoor settings in wide open air environments with sunlight as the light source to activate the photocatalytic surfaces.

5.1. Indoor DeNOxing Environment Tests with TiO₂ Photocatalytic Cementitious Surfaces

One of the first articles in the search for means to reduce NO_x from the various emission sources, which affected the air quality in Japan because of relatively high concentrations of NO_x, particularly along heavily trafficked roads in densely populated areas, was that of Ibusuki and Takeuchi in 1994 [94]. They proposed a new non-thermal technology for removing NO_x at sub-ppm level from the air of trafficked roads, tunnels and other environmental emission sources (see Section 4) following their earlier findings [91] that NO_x deposited onto the ground caused an enhancement of the rate of adsorption of NO_x on soil particles upon photoillumination, with TiO₂ exhibiting the highest activity for NO_x adsorption—about 60% of NO_x was captured as nitric acid on the TiO₂ surface. Their follow-up laboratory experiments indicated that up to 90% of NO could be removed using mixtures of TiO₂, activated carbon (AC), and iron oxide (Fe₂O₃) particulates in a flow-through reactor [96].

Along similar lines, at the 2000 JIPEA World Congress, Murata and coworkers [122] reported the development of interlocking cementitious bricks loaded with TiO₂ (NOXER^©) for the oxidative removal of NO_x under sunlight UV radiation, humidity, and NO_x concentrations that paralleled roadside environments. This novel technology at the time was implemented in a couple of Tokyo suburbs. The permeable interlocking bricks (Figure 30) were prepared by mixing aggregates, cement, TiO₂ powder and water in an appropriate template, and then cured at ambient temperature for one month, after which they were used for the photocatalytic indoor tests in a flow-through small PVC reactor (conditions in Figure 31).

After 12 h of UV irradiation, ca. 88% of NO was removed at a relative humidity (RH) of 10%, decreasing to 52% at 80% RH, whereas the quantity of NO removed increased with increase in UV intensity (from 10% at 0 W m⁻² to 88% at 12 W m⁻²). Varying the NO concentration from 0.05 to 5.0 ppm showed that in the 0.05–0.15 ppm range of NO—a range similar to roadside levels—a constant removal of ca 90% was observed decreasing to ca. 45% at 5 ppm of NO [122]. In an outdoor test in which NO_x from the roadside (ca. 0.5 ppm) was passed through the PVC reactor exposed to natural sunlight (T = 17 °C and RH = 47%) led to ca. 80% decrease in NO_x.

Evidence of the effectiveness of NO_x abatement was also carried out by the Italcementi Group using their TX-Active^® photocatalytic TiO₂ deposited on a suitable substrate [121,123] and subsequently placed in a small reactor (top in Figure 32 [121]) of predetermined volume in which NO_x gas was introduced, and then diluted with air to reach a certain pollutant concentration; the schematic of the flow-through reactor assembly is also displayed (bottom of Figure 32). The efficacy against NO_x gases was also demonstrated during the project PICADA with tests conducted at the European Laboratory of Ispra (Italy), inside an Indoortron—an environmental chamber with people access and characterized by such controlled parameters as temperature, relative humidity, air quality and air exchange rate—to also study the fate of various other internal VOC contaminants. Figure 33 illustrates the time course of the removal of NO_x with the reactor assembly of Figure 32. Evidently, complete NO_x removal under the ISO conditions occurred within ca. 6 h of light (UV) irradiation in the presence of TX Active^® in the absence of which no changes in NO_x concentration occurred. Tests on the best formulations of a white and a grey photocatalytic paint with TX Active^®, chosen for tunnel renovation (see below), showed a NO_x abatement capacity of 88–90% after only 60 min under UV light irradiation in the reactor assembly of Figure 32; however, under similar conditions, the same paints necessitated nearly 26 h of UV irradiation to decolorize 70% of a rhodamine-B dye stain on the paint’s surface [124].

An otherwise similar reactor assembly was used by Martinez and coworkers [125] to examine the degradation of NO present in the air by means of a photocatalytic oxidation process based on TiO₂ nanoparticles that had been incorporated in a polymer-matrix-based coating. The experimental setup consisted of a flow type reactor (Figure 34) adapted from the ISO 22197-1 standard; the final products detected were NO₂ in the gas phase and nitrate ions adsorbed on the photocatalytic surface. The photocatalyst (anatase-TiO₂) was a commercial slurry solution available from Evonik (Aerodisp^® W740X). The coatings were primarily formulated for the surface treatment of building materials. To identify the possible influence of the nature of the substrate on the photocatalytic efficiency, various types of substrates were tested. The coatings were applied to: (i) mortars; (ii) glass plates; and (iii) non-absorbent cardboard materials. The wet thickness of coatings was 40 μm (other conditions are reported in Figure 34).

Results indicated that irradiation for 60 min and relative humidity of 60% led to ca. 25% of an initial concentration of 8.61 μmol of NO to be degraded at a flow rate of 1.5 L min⁻¹ [125]. On mortar and glass substrates, the influence of increasing humidity on the degradation rates depended on the nature of the substrate and on initial NO concentrations; no significant influence of humidity was observed at initial NO concentrations of 400 and 1000 ppb, while a significant decrease in the kinetics was seen with a decrease of humidity at higher initial concentrations of NO (1500 and 2000 ppb). Generation of NO₂ on mortar was very low, because of good adsorption capacities of the supporting substrate. On glass, NO_x degradation rates decreased strongly on generation of NO₂ owing to competition between pollutant and humidity (water) for the adsorption sites [125].

Using a specifically-developed test apparatus (Figure 35), Staub de Melo and Triches [126] assessed the efficiency of a photocatalytic mortar under no less than 27 different environmental conditions: varying the relative air humidity (30%, 50% and 70%), the UVA radiation (10, 25 and 40 W m⁻²), pollutant mass flow rate (1, 3 and 5 L min⁻¹), and initial concentration of NO (20 ppmv). Results showed that the higher were the levels of UVA radiation, the better was the performance of the mortar in degrading NO_x. By contrast, at higher relative humidity levels and flow rate caused a decrease in photocatalytic activity, which showed that environmental conditions have a significant impact on the efficiency of the photocatalytic mortar in the degradation of NO_x.

The photocatalytic mortar was produced using Portland cement with Pozzolan (CP II Z 32); the catalyst was a nanometric rutile TiO₂ bar (dia. = 10 nm; length = 40 nm; 98% purity; specific surface area = 150 ± 10 m² g⁻¹; real density = 4.23 g cm⁻³). Mortars with addition of 3%, 6% and 10% TiO₂ were investigated; layers of photocatalytic mortar with thicknesses of 3, 6 and 10 mm were applied to samples of precast concrete paving (PCP). The gas system consisted of dry air and 500 ppmv of NO stabilized with N₂ gas, which simulated a polluted atmosphere. Maximal removal of NO_x was 50% for an initial 20 ppmv of NO, relative humidity of 50%, a flow rate of 1.0 L min⁻¹ and a UVA irradiation (10 W m⁻²) period of 25 min. A cost to photocatalytic efficiency evaluation led to a 3-mm coated mortar incorporating 3% of TiO₂. For the application of such materials, the authors [126] suggested that locations with lower relative humidity, high incidence solar radiation and little air mass movement should be sought in the field, as they would provide better conditions to achieve high efficiency of the TiO₂-coated precast concrete paving materials.

An examination of the past literature shows that different types of photoreactors have been used in standardization methods to quantify the activity of photocatalysts in air remediation with commercially available photocatalytic materials. Classically, the degradation of NO_x has been a major subject of investigations because relatively simple and inexpensive chemiluminescence instruments are available to quantify NO_x and because NO₂ is of crucial importance for urban air quality. Nonetheless, when investigating the photodegradation of NO_x in laboratory settings, only NO was used because of its facile detection, its lower ability to adsorb on reactor surfaces, and because of slower dark reactions that might occur on photocatalytic surfaces [127]. To reduce the time for establishing adsorption equilibrium and to increase precision of the NO_x data from low-sensitive instruments, unrealistically high NO concentrations (500–1000 ppbv) have been commonly used; and as analyses of reaction products require more sophisticated instrumentation, no reaction products (e.g., nitrite and nitrate) other than NO₂ are quantified. To the extent that NO₂ is an intermediate in the photocatalytic oxidation of NO by reaction with O₂^−•/HOO^• radicals, the photocatalytic removal of NO_x, not just NO, is quantified by different standardization methods that are still under development (as we speak).

In this regard, Ifang and coworkers [127] demonstrated that transport limitations can lead to an underestimation of the activity, if fast heterogeneous reactions were investigated in bed photoreactors. When using stirred tank photoreactors, complex secondary chemistry can lead to an overestimation of the photocatalytic remediation of NO_x, if NO₂ were also present, not to mention that the quantities used for ranking the activity of photocatalysts in air remediation in the different methods currently used are not independent of experimental conditions, so that any inter- comparison between different methods or extrapolation to atmospheric conditions is a futile exercise. Consequently, the authors [127] proposed a modified method for quantifying air remediation activity of photocatalytic surfaces that would overcome such problems. The method is based on a bed flow reactor (Figure 36) that can easily be adapted to the ISO method. The extent of degradation of NO_x on photocatalytic surfaces in continuous stirred tank reactors can be significantly influenced: (a) by the gas-phase Leigthon chemistry (Equations (85)–(87)); (b) by unwanted wall losses of reactive agents and products; and (c) by heterogeneous formation of products on reactor walls (e.g., HONO) [127]. In such a photoreactor, short reaction times of only a few seconds and more homogeneous inert surfaces (no fan, among others) should minimize the aforementioned issues.

NO₂ + hν → NO + O(³P)

(85)

O(³P) + O₂ → O₃

(86)

NO + O₃ → NO₂ + O₂

(87)

The bed flow photoreactor was constructed of a single block of Teflon in which photocatalytic samples up to 40 cm × 5 cm × 1 cm could be investigated; recommended light sources were two 20-W UVA fluorescence lamps (300–500 nm, λ_max = 370 nm; length, 57 cm) mounted at variable distances to the reactor to adjust the irradiance level measured by a calibrated spectroradiometer [127]. Because of differences in photoactivity of commercial photocatalysts toward NO_x (and VOCs), the authors suggested that at least one compound from each class be examined in standardization methods for air remediation. Moreover, to avoid saturation problems, laboratory experiments would have to be performed under relevant atmospheric conditions (i.e., for RH = 50%; reactant concentrations ≤ 100 ppb), and, as NO₂ is of much greater environmental importance compared to NO, the use of NO₂ as the test reactant was strongly recommended [127].

On their part, Zouzelka and Rathousky [128] investigated the photocatalytic activity of two commercial titania-based products: (1) Protectam FN2 that consisted of ca. 74% of Evonik’s Aeroxide P-25 TiO₂ powder and 26% of an inorganic binder; and (2) Aeroxide P-25 TiO₂ powder as photocatalytic coatings (10 μm thick) on concrete and plaster supports toward the abatement of NO and NO₂. Photocatalytic experiments on the coatings were performed in two types of flow reactors, one with laminar flow while the other with an ideally-mixed flow (Figure 37), under real world conditions in terms of temperature, relative humidity, irradiation intensity and pollutant concentrations. Results showed that the photocatalytic process reduced significantly the concentration of both NO_x agents in the air.

The decrease in the concentration of NO_x achieved in the steady-state for an inlet concentration of NO and NO₂ of 0.1 ppmv, corresponding to highly polluted urban air, was up to 75 μmol m⁻² h⁻¹ (for the Protectam FN2) and 50 μmol m⁻² h⁻¹ (for the Aeroxide P-25 TiO₂) at a flow rate of 3000 cm³ min⁻¹ and a relative humidity of 50%. Because of a conspicuous lack of data regarding the performance of photocatalytic coatings over long periods, the authors [128] also examined aged photocatalytic FN2 coatings on a 300-m² concrete noise barrier that had been exposed to heavily-trafficked (ca. 30,000 vehicles a day) thoroughfare in Prague where the NO_x concentration reached 30–40 μg m⁻³, often exceeding the permitted NO₂ limit of 40 μg m⁻³ (or 0.021 ppmv).

Experimentally, the area of irradiated photocatalytic surface was 50 cm² (5 cm × 10 cm); flow rate of air mixture was 3000 cm³ min⁻¹; total volume of air treated in 24 h was 4.32 × 10⁶ cm³; volume of purified air and area of irradiated photocatalytic surface were the same in both photoreactors, although the reactors differed substantially in volume/irradiated area (65 times greater for the ideally-mixed flow reactor); volume of the ideally-mixing flow reactor was 5200 cm³ (18 × 32 × 9 cm); the free volume of the laminar flow reactor was 80 cm³ (5 × 32 × 0.5 cm); and linear streaming velocity of the gas was 0.2 m s⁻¹. The bandgaps of the TiO₂ in the two materials were 3.2 eV for the P-25 sample and 3.05 eV for the FN2 sample, the red-shift in the latter being attributed to the effect of the binder on titania [128].

Comparison of the photocatalytic performance of P-25 and FN2 materials reported in Figure 38 shows the reaction rate with the FN2 coating to be greater than for the P-25, even though the quantity of TiO₂ in the FN2 coating was lower. For instance, the reaction rate with the FN2 coating on concrete in both laminar and ideally-mixed flow reactors at an inlet NO concentration of 1.0 ppmv was 40% and 49% higher, respectively, than the corresponding reaction rate of P-25 which, according to the authors [128], was likely due to the nearly twofold larger surface area of the FN2 specimen (82 m² g⁻¹) relative to the P-25 sample (47 m² g⁻¹).

Evidently, the inlet concentration of NO had a substantial influence on the reaction rate in both reactors, as evidenced on comparing the data for the FN2 coating in the laminar flow reactor in which the reaction rate was approximately proportional to the inlet NO concentration; in other words, for an inlet NO concentration of 1.0 ppmv, the rate was about ten times greater than for a concentration of 0.1 ppmv [128]. With regard to the effect of NO₂ concentration used (0.1 ppmv) the trends differed from those for NO. First, the difference in the reaction rate at the beginning of the reaction and in the steady-state was much smaller for both P-25 and FN2 coatings when applied on plaster than on concrete. On plaster, the reaction rate was practically unchanged, whereas on concrete the decrease in rate was between 40% and 60%. Moreover, the steady-state reaction rate (after a 24 h period) on the coatings applied on plaster was consistently 1.5–1.8 times higher in comparison with the coatings on concrete.

Another important issue emphasized by Zouzelka and Rathousky [128] was the durability of the performance of the photocatalytic coatings under real-world conditions, as exemplified by the commercial photocatalytic coating Protectam FN2 that maintained relatively high efficiency in removing NO_x from contaminated air even after two years under the harsh conditions noted above; this was likely due to good mechanical properties of the binder.

Moving from small benchtop laboratory photoreactors to actual field studies to investigate the efficiencies (activities) of photocatalytic concrete/mortar with embedded titania photocatalysts, simulation chambers have also proven suitable to test photocatalytic materials under controlled environmental conditions, as they represent a mid-way step between laboratory and actual environ-ment (a sort of pilot plant scale). Additionally, simulation chambers offer the opportunity to examine the impact of photocatalytic surfaces, also on secondary air chemistry, in contrast to the smaller flow-type reactors. In this regard, within the framework of the Life + project PhotoPAQ (2010–2014) that will be described later (Section 5.2), Mothes and coworkers [25] tested the behavior of O₃, NO_x, and selected VOCs (toluene and isoprene) on a photocatalytic cementitious coating material (with Italcementi’s TX-Active^®) under UV irradiation and atmospherically relevant conditions (relative humidity, temperature, and realistic pollutant concentration) using the aerosol chamber LEipziger Aerosol Kammer (LEAK) at the Leibniz Institute for Tropospheric Research (TROPOS)—a 19 m³ aerosol cylindrical chamber with a surface-to-volume ratio of 2.0 m⁻¹ and illuminated with UV lamps (λ = 300–400 nm) with an average light intensity of 11 W m⁻².

The TX-Active^® material was mixed with ultrapure water and subsequently applied manually on both sides of sand-blasted glass plates supported by a home-made device to obtain a rough but uniform surface (final thickness, ca. 3–4 mm), after which the material was cured for ca. 1 month. The cleaned coated glass plates were then inserted into a special aluminum rack installed inside the chamber—the area of the photocatalytically active surface was 6.65 m² (boosted specimen)—a blank specimen devoid of TX-Active^® titania was also made for comparison. Both specimens were flushed with clean air (200 L min⁻¹) with the UV lamps turned ON for ca. 30 h to remove any potentially adsorbed pollutants; RH varied between dry air and 50%—the photocatalytic tests were performed following the 2007 ISO-22197-1 standard methodology. Results obtained for the NO and NO_x degradation experiments together with formation of O₃ at three RHs and at 293 K are displayed in Figure 39; dark period ca. 120 min, irradiation period also 120 min, initial concentrations of NO and NO_x = 40 ppb [25].

Contrary to the behavior of NO_x during the dark period, NO showed only a slightly different behavior at the different RHs; NO’s behavior was nearly constant at 50% RH and 70% RH, whereas at 10% RH a slight decrease was seen that was attributed to slightly adsorbed NO on the specimens’ surface [25]. Comparison of the blank and boosted materials revealed that UV irradiation had a significant influence on NO_x owing to the photocatalytic properties of TiO₂ embedded in the boosted material. The small increase of NO at the beginning of the light period was attributed by the authors [25] to photolysis of NO₂ according to the Leighton sequence described by Reactions (85)–(87).

Results from the chamber studies also showed a photocatalytic effect on O₃, evidenced by an increase of the geometric uptake coefficient from 5.2 × 10⁻⁶ for the inactive to 7.7 × 10⁻⁶ for the active material under irradiation. Measured first-order rate constants for NO_x under irradiation ranged from 2.6 × 10⁻⁴ s⁻¹ to 5.9 × 10⁻⁴ s⁻¹, significantly higher compared to the range for the inactive materials (7.3–9.7 × 10⁻⁵ s⁻¹), thus demonstrating the photocatalytic effect. However, no significant photo- catalytic degradation was observed for the VOCs (toluene, isoprene); the upper limit uptake coefficient for both was only 5.0 × 10⁻⁷. Small carbonyl (C1–C5) gas-phase compounds were identified when using the photocatalytically active material, a result of the photocatalytic degradation of the organic additives [25]. In contrast to the uptake observed for pure O₃, a clear photocatalytic formation of O₃ was observed (k_o(O₃) ≈ 5 × 10⁷ molecules cm⁻³ s⁻¹) during the experiments with NO_x (RH ≥ 50%). The authors further noted the necessity for detailed studies of heterogeneous reactions on such surfaces under more complex simulated atmospheric conditions as enabled by simulation chambers.

Although many studies have been done on TiO₂ photocatalytic cementitious substrates to deNOxify the environment—as we discuss below—nearly all studies neglected to consider the inter- relationship between the metal-oxide photocatalyst and the non-negligible chemistry that might occur on the cementitious support—in fact, this inter-relationship is particularly relevant for an appropriate understanding of deNOxification processes. In this regard, the investigations carried out by Macphee and Folli [129] on Photocatalytic concretes—The interface between photocatalysis and cement chemistry have opened up the proverbial Pandora’s Box in which they addressed: (i) the photocatalytic mechanisms applicable to atmospheric depollution; (ii) the influence of doping the metal oxide; and (iii) the application of TiO₂-based photocatalysts to concrete. These authors further emphasized some of the points described earlier that the catalyst efficiency is influenced by several factors, none of which are negligible: for instance: (i) energy and intensity of the activating radiation incident on the photocatalyst; (ii) the number and relative energy positions of the electronic states in the photocatalyst, defined by the crystal structure and the redox potentials associated with the required redox processes; (iii) the charge carrier mobility within the semiconductor’s conduction and valence bands; (iv) the kinetics of charge transfer processes; and (v) the accessible catalyst surface for adsorption of the pollutants, oxygen, and water [129].

Of greater importance, they noted that the chemistry of the cement environment is quite different from the ambient conditions normally prevalent in environmental photocatalysis, which will have a non-insignificant influence on the normal behavior of the catalysts in the concrete [129]. In freshly mixed cement, the high pH and the high ionic strength aqueous mix containing multiply charged ions dramatically modifies the surface chemistry and behavior of TiO₂ dispersions, and thus will have a strong impact on dispersion behavior and adsorption properties, as the mode of adsorption and strength of binding between an adsorbing molecule and a surface are conditioned by their relative charges. Additionally, pH changes during the early stages of mixing and ageing the cementitious substrate will also have an impact on the band edge positions in the semiconductor photocatalyst as well as the oxidation chemistry of NO_x. The alkalinity of wet cement/concrete is typically greater than pH 13, which can change rapidly at the surface as the concrete sets and the formwork is removed (Figure 40a) [129]. The highly alkaline aqueous phase can react with atmospheric CO₂ to produce CO₃²⁻ ions, causing the pH to decrease and induces surface deposition of calcite—a carbonate mineral and the most stable polymorph of calcium carbonate—that can obstruct deposited photocatalysts. Figure 40b–d displays the microscopic details of the surface calcite layer that obstructs TiO₂ clusters present in the hardened cementitious matrix.

The alkali nature of the cementitious surface also has significant consequences on the chemistry taking place at the surface when NO_x molecules are adsorbed, even in the absence of any catalyst and under non-irradiating conditions, as both NO and NO₂ (NO_x molecules) can undergo an oxidative reaction with the alkali to yield nitrite and nitrate ions (Equations (88) and (89)) [130]. These two reactions must be considered whenever photocatalytic efficiencies are assessed in deNOxification processes.

2 NO₂ + 2 OH⁻ → NO₂⁻ + NO₃⁻ + H₂O

(88)

NO₂ + NO + 2 OH⁻ → 2 NO₂⁻ + H₂O

(89)

To address the geographical limitations of UV intensity and recognizing that sunlight is the light source to activate the photocatalyst in practical applications of photocatalytic concrete in deNOxification, the Macphee group [129,131] also investigated M-doped titania (M = W, Nb) photo-catalysts toward removal of NO_x because unmodified TiO₂ releases a significant quantity of toxic NO₂ in the deNOxification process, a problem rarely raised in several previous studies as NO_x abatement is commonly assessed only on the basis of NO removal rather than NO₂ or total NO_x removal. Accordingly, to highlight the problem, they performed a systematic study on a number of commercial titania powders and doped titania specimens by determining both the photocatalytic activity (i.e., the photonic efficiency ξ [29], as defined by Equation (90)) and the selectivity (S defined by Equation (91)) toward formation of NO₂ and NO₃⁻ in the oxidation of NO.

$$\xi =\frac{\left(C_{dark}-C_{irr}\right)Vp}{{\rho }_{o}ART}$$

(90)

$$S=\frac{{\xi }_{N{O}_x}}{{\xi }_{NO}}$$

(91)

where C_dark and C_irr are the concentrations of the species of interest under dark and irradiated conditions, respectively; V is the volumetric flow rate; p is the pressure; ρ_o is the photon flux impinging on the photocatalyst surface; A is the irradiated area; R the gas constant; and T is the Kelvin temperature.

In the case of W-doped TiO₂, Figure 41 demonstrates that while doping titania with tungsten significantly improved the selectivity toward formation of NO₃⁻, it considerably reduced the photonic efficiency [131]. This raised the question as to how one could evaluate various photocatalysts toward NO_x abatement when both activity and selectivity needed to be considered. To resolve this matter, the authors [129,131] defined a new figure of merit: the DeNO_x index (Equation (92)), which was derived by assigning a toxicity value to both NO and NO₂ and then expressing the change in total toxicity rather than the concentration change of the individual NO_x gases [129,131]:

$${\xi }_{deN{O}_x}={\xi }_{NO}-3{\xi }_{N{O}_2}$$

(92)

The deNO_x index is positive if the photocatalyst lowers the NO_x toxicity.

Figure 42a illustrates the results from several photocatalyst compositions and demonstrates a positive deNO_x index at nominal W contents > 4.2 at.% for the W-doped TiO₂ [129]. A range of commercial TiO₂ photocatalysts were tested under comparable conditions giving a DeNO_x index between 0 and −4000 representing poor nitrate selectivity and, most importantly, meaning that the more toxic NO₂ was generated [129]. These data were further supported by analyses on powder and mortar samples incorporating W-doped and Nb-doped TiO₂, as displayed in Figure 42b [129]. Interestingly, although undoped P-25 TiO₂ has been the workhorse in photocatalysis, it displayed a net negative effect owing to its high activity but low nitrate selectivity; that is, it efficiently converted NO to NO₂ rather than to NO₃⁻ ions. By contrast, both W- and Nb-doped TiO₂ showed a higher tendency towards NO₃⁻ and thus are better suited as remediation photocatalysts, whether in powdered form or embedded in mortars.

Continuing their interesting studies on photocatalytic concrete, in their most recent article, the Macphee group [132] examined the effect of photocatalyst placement as regards photocatalyst efficiencies in concrete technology, and pointed out that even though the technology represents a well-established concept and notwithstanding the significant opportunities for air quality improve-ments to be derived from the considerable concrete surfaces exposed to the atmosphere, especially in cities highly polluted by vehicle exhaust and industrial emissions, photocatalytic concretes have so far remained in the investigative sphere, rather than in the mainstream of applications. As with any commercial new technology, the likely barriers for widespread implementation may well be cost effectiveness that might emanate from the photocatalyst impact and the challenges in measuring directly what the impact of this new concrete technology may actually be on air quality. The challenges are indeed very complex. Accordingly, the authors [132] placed photocatalytic efficiencies into context by comparing performances of conventional photo-catalyst dispersions in surface mortar coatings vis-à-vis photocatalysts supported on surface-exposed aggregates as well as on the nature and impact of catalyst binding to aggregate supports. However, as we discuss below, the efficiencies in real-world environmental applications differ significantly from the efficiencies experienced in a laboratory setting where small photo-reactors have been used to measure the efficacies of various TiO₂ photocatalyst concrete specimens indoor, wherein experimental conditions can be controlled in contrast to outdoor environmental conditions, which are not only widely different, but are also uncontrollable conditions that one experiences in the real world. Nonetheless, their findings in photocatalyst placement as regards efficacies of photocatalytic concrete are worth noting [132].

In treating NO_x gases by TiO₂-based photocatalytic substrates, the currently accepted sequence is represented by Reaction (93); the greater is the degree of conversion of NO and NO₂ to nitrates, the greater is the catalytic activity of the substrates.

NO → HNO₂ → NO₂ → NO₃⁻

(93)

Although this oxidation sequence is thought to be the preferred reaction with respect to air quality, as claimed by the authors [132], a far better process to improve air quality would be the reduction of NO_x back to N₂ and O₂ as can be achieved by selective catalytic reduction (SCR) and selective non-catalytic reduction (SNCR) technologies (see above). Regardless, in TiO₂-based photocatalytic methods, both oxidative entities (valence band holes) and reductive entities (conduction band electrons) are generated upon sunlight UV activation of the TiO₂ photocatalyst, so that the oxidation sequence (Reaction (93)) may also lead to intermediates—for example, ozone and peroxyacyl nitrates—in addition to those in the sequence (Reactions (94)–(96)) produced in surface processes in the NO₂ production regime, and those produced in the sequence (Reactions (97)–(100)) in surface processes in the N₂O₅ production regime when ozone is present [133], all of which could be released into the atmospheric environment not forgetting that NO₂ is nearly three times more toxic than NO.

2 NO_2(ads) ←→ NO₃⁻_(ads) + NO⁺_(ads)

(94)

NO⁺_(ads) + O²⁻_(ads) ←→ NO₂⁻_(ads)

(95)

NO₂⁻_(ads) + NO_2(ads) ←→ NO₃⁻_(ads) + NO_(gas)

(96)

TiO₂ + O₃ → O–TiO₂ + O₂

(97)

NO_2(ads) + O–TiO₂ → NO₃ − TiO₂

(98)

NO_3(ads) + NO_2(g) → N₂O_5(ads)

(99)

N₂O_5(ads) → N₂O_5(g)

(100)

Notwithstanding the above sequences, accumulation of NO₃⁻ on catalytic surfaces must be managed since adsorbed NO₃⁻ ions block catalytic sites from further NO/NO₂ adsorption—thus, affecting catalytic activity—as the nitrates back react to form NO₂ thereby causing nitrate selectivity to be diminished. This calls attention to the notion that reNOxification may be an important event in deNOxification processes. Accordingly, NO₃⁻ must be removed periodically by washings [132] at least on a weekly basis for certain infrastructures (tunnels, noise barriers, and buildings, among others) but especially for road tunnels where rain has no impact; this maintenance would carry increased costs. Moreover, where rain is sparse in the summer months as often occurs in Southern Italy and California (USA), such periodic washings may well be necessary. In addition, in conventional photo-catalytic concretes, nitrate washings via condensed atmospheric moisture represents but a mild risk from acidification of the cement (carbonated at the near surface) which may influence leaching-induced damage over time [132].

In their earlier articles [129,131], Macphee’s group addressed various factors that limit photocatalytic efficiencies in cement-based systems in which the metal-oxide photocatalyst was incorporated in the mortar/concrete. They have now proposed to separate the chemistry taking place on the photocatalyst surface from the chemistry occurring on the cementitious support, so as to enhance photocatalytic efficiency by depositing the photocatalyst on the surface of the cementitious aggregate (Figure 43) [132]; that would require some sort of binder which unfortunately would also add to the complexity of the events as the binder may not only photodegrade but may also bring about changes to the electronic structure of the photocatalyst [128].

The aggregate used to support the prepared TiO₂ photocatalyst was quartz sand (1–2 mm diameter) that was subsequently treated in aqueous solution of Ca(OH)₂ followed by carbonation with CO₂ (aggregates denoted QST) [132]. The aggregates were placed in the flow-through reactor displayed in Figure 44 (monolayer; area, 5 × 10⁻³ m²), after which they were irradiated with a 500-W Xe-lamp solar illuminator (photon flux at the aggregates, 3.05 μmol m⁻² s⁻¹); other conditions were: 1 ppm NO gas in synthetic air; relative humidity, 40%; temperature, 25 °C; volumetric flow rate, 5 × 10⁻⁵ m³ s⁻¹ through the reactor.

Three stages were identified by the authors in the process of converting NO to nitrate: (i) stabilization of NO concentrations on the QST specimens in the dark for achieving an adsorption/desorption equilibrium; (ii) the photocatalytic conversion of NO under illumination; and (iii) recovery of NO concentrations after illumination was terminated [132]. Results showed that the concentration of NO decreased rapidly on illumination, remaining nearly constant for a while, whereas the concentration of NO₂ increased significantly during this time; stopping the illumination caused the NO₂ concentration to go back to zero (Figure 45). The photonic efficiency ξ_NO increased initially with increasing TiO₂ mass fractions but became steady around 0.76%. Concomitantly, the ξ_NOx and nitrate selectivity (S%) increased with increasing mass fraction of TiO₂, which led to a corresponding decrease of the rate of production of NO₂. Compared with the NO_x removal results for pure TiO₂, the estimated ξ_NO for the QST specimens was ca. 0.6%, while nitrate selectivity was ca. 40%; the supported TiO₂ particles presented higher photonic efficiency for NO removal, but lower nitrate selectivity owing to reduced NO₂ removal efficiency.

Clearly, the photocatalytic performance depended on the TiO₂ loading with the activity increasing with TiO₂ content up to ca. 0.76%, while nitrate selectivity increased across the range of TiO₂ loading used, and was strongly conditioned by the negative influence of Ti–O–Si bonds on selectivity and the effect of dilution as TiO₂ loading was increased. Washing to remove weakly bonded TiO₂ led to a decrease of the photocatalytic activity ξ_NO by about 30%, even though nearly half of the TiO₂ had been lost (only 0.34% remaining); however, selectivity was not significantly affected [132]. While such a low photocatalytic activity could be explained, in part, by fast charge carrier recombination and charge carrier trapping dynamics within intra-bandgap defect states in TiO₂ photocatalytic substrates, mechanical loss of titania particulates from the support, coupled with the observed activity under controlled indoor experimental conditions (small flow-through reactor), as reported by the authors [132], do not bode well to convince implementation of the photocatalytic concrete technology to outdoor real-world environments where, as noted earlier, the conditions are different and uncontrollable.

The results of the few studies reported herein with regard to the extent of removal of NO_x gases (NO and NO₂) are summarized in

Table 7. Percent removal of NO_x gases carried out with TiO₂-based photocatalytic protocols in an indoor laboratory setting using small flow-through reactors under a variety of conditions.

Study	Gas	TiO2-Based Photocatalytic Removal of NOx Gases	Year Study Was Carried Out	Refs
1	NOx	60–90%	1994	[97]
2	NO	52–88%	2000	[122]
		45–90%
3	NOx	88–90%	2009, 2011	[121,124]
4	NO	25%	2011	[125]
5	NOx	50%	2012	[126]
6	NOx	40–60%	2017	[128]
7	NOx	35%; 66%	2016	[25]
8	NO	44%	2018	[132]
	NOx	16%
9	NO	80%	2006	[134]

[25,97,121,122,124,125,126,128,132,134] for indoor laboratory experiments carried out with small bed flow-through reactors employing a variety of different conditions from one study to the next. Various standardization protocols have been proposed in the last few years that have led to significant variation in the results—as should have been expected.

5.2. DeNOxing the OUTDOOR Environment with TiO₂ Photocatalytic Cementitious Materials

As shown above, many studies have been reported, albeit not all have been cited herein, on the photocatalytic abatement of NO_x in a laboratory setting wherein experimental conditions could be controlled. Relatively speaking, however, only a few field trials have been conducted outdoors in open air environment under real-world conditions; as we have noted, the conditions are not only different but, more importantly, they also cannot be controlled. As an example of the latter, some reduction of NO_x levels was achieved at the laboratory scale and in outdoor field experiments in Antwerp (Belgium) where a surface of ca. 10,000 m² was covered with photocatalytic pavement blocks—researchers that did the study were unable to conclude the extent to which NO_x was reduced because the measurement period was too short [135]. Additionally, an experiment to test the efficacy of photocatalytic TiO₂-mortar panels to degrade NO_x in a setting that involved artificial canyon streets showed NO_x concentrations to be reduced by as much as 37–82% depending on the conditions [136]. Likewise, a study carried out in 2006 in Bergamo (Italy) over a four-week period in a segment of a local street covered with photocatalytic paving stones showed a 30–40% reduction in the NO_x concentration compared to a similar section left untreated [137]; elsewhere, NO_x reduction levels of 20 to 50% have been quoted that depended on weather and traffic conditions [138]. A study done in the city of Segrate (Italy) [137] where a concrete road segment (7000 m²) was constructed using a thin-layer of a photocatalytic mortar was said to display a 57% NO_x abatement level, while a similar study in an industrial section near the town of Calusco (Italy) reported a NO_x abatement level of 45% for an 8000 m² pavement built with photocatalytic concrete blocks. By contrast, field studies carried out in an artificial model street canyon in Petosino, Italy [15] and in the Leopold II tunnel in Brussels (Belgium) [14] reported photocatalytic reductions of NO_x well below measurement precision errors (1–2%). An “indoor” field study of the Umberto I Tunnel (Rome), for which the walls were coated with a photocatalytic paint and a new UV-Visible lighting system installed, showed more than 20% reduction of NO_x [138]. The latter two field studies [14,138] are discussed in some detail below.

Clearly, the many discrepancies in NO_x reduction levels when using photocatalytic paints, photocatalytic paving bricks, or photocatalytic mortar/concrete on streets, highways, or roadside noise barriers: (i) call attention to highly differing prevailing atmospheric conditions in field experiments; (ii) call into question how these field experiments were carried out; and (iii) call into question how the empirical results were interpreted. For that reason, we now describe the various attempts carried out using TiO₂-based photocatalytic substrates/aggregates in a few countries and discuss their findings with an emphasis on three major undertakings funded wholly or in part by the European Union: the PICADA Project (2002–2006); the Life⁺-funded Project PhotoPAQ (2010–2014); and the LIFE MINOX-STREET Project (2014–2018).

One of the first applications of photocatalytic cement-based materials for self-cleaning purposes was Richard Meier’s Dives in Misericordia Church project in Rome where Italcementi served as the principal technical sponsor, analyzed the concrete mix (Figure 46) [139], and provided continuous supervision during the construction period (1999–2001). The church was erected with three huge, totally white sails made of precast photocatalytic concrete blocks (Figure 47) that would ensure unparalleled and time-enduring white color to the built elements thanks to the self-cleaning properties of the final TX-Active^® surfaces [140,141]. The photocatalytic cement-based product used was expected to maintain its aesthetic characteristics unchanged over time, especially the color, even in the presence of aggressive urban environments. Laboratory pilot-scale tests demonstrated that NO_x abatement with photocatalytic cement-based products was impacted by temperature, relative humidity, contact time of NO_x with the surface (flow velocity, height of the air flow above the sample, among others); it was also noted that reduction efficiency increased with longer contact times (larger surfaces, lower velocities, and higher turbulence/mixing), higher temperatures and lower relative humidity—however, in real situations, conditions such as high temperatures, no wind and no rain present the largest risk of ozone formation [142].

Guerrini and Corazza [142] claimed that a building element containing cement to which TiO₂ had been added was capable of maintaining its aesthetic appearance unaltered in time (see also [121]), thus contributing to reduce the dirtiness of surfaces exposed to specific polluted environments. A look at the photographs illustrated in Figure 48, however, shows that the long-term effectiveness of anatase TiO₂ photocatalytic cement-based external walls of the church failed the test of time.

Another important sponsorship of the CTG-Italcementi Group was the renovation of the Umberto I Tunnel located in the center of Rome (dimensions: length, 347.7 m; width, 17 m; height: 8.5 m) for which renovation, carried out in the Summer of 2007, was deemed necessary because, among others, of the griminess of the indoor vault that was coated with an oily thickness of smog (Figure 49a) [123]. After a thorough clean-up, the tunnel walls were coated with a photocatalytic cement-based paint, with the gray paint applied on both sides of the tunnel (up to 1.80 m height from the road surface) and with a white paint for the remaining surface (total surface, 9000 m²)—it is not clear, however, what the source of the paints was (Cimax Ecosystem Paint by Calci Idrate Marcellina srl, a photovoltaic product patented by Italcementi [123], or Airlite paint, as claimed by Ref. [143]). Pollution and weather conditions were expected to be less variable in the indoor vault of the tunnel than in a typical outdoor environment, thus the researchers thought would facilitate the evaluation of the photocatalytic depolluting action of the photocatalytic grey and white paints with a reduced number of parameters and making the interpretation of the data less complex. During the first campaign of July 2007, the extent of NO_x abatement was determined to be 20–25% in the center of the tunnel, whereas in the second campaign of September–October 2007, the NO_x abatement level was estimated at 51–64% based on a statistical approach [138]. Regardless, Figure 49b [123] displays a photograph of the renovated tunnel before inauguration day of 9 September 2007, while the photographs in Figure 50 show the characteristics of the Umberto I Tunnel in Rome nearly 11 years after the renovation. Evidently, the long-term effectiveness of the photocatalytic product is also an issue here, not to mention the degradation (blisters) of the painted cementitious walls.

As the adsorbed nitrates (NO₃⁻) formed subsequent to the oxidative conversion of the NO_x gases on the surface of the photocatalytic cement-based paints cannot be washed off from the tunnel’s indoor walls, a problem arises with regard to the chemistry that occurs on the walls as accumulation of adsorbed NO₃⁻ on the catalytic surfaces block the catalytic sites from further NO/NO₂ adsorption—thus, not only is the catalytic activity of the photocatalytic paint affected, but also the NO₃⁻ may be reduced catalytically back to NO₂. In other words, reNOxification is likely to occur subsequent to the deNOxification process, unless the adsorbed nitrates are removed regularly by washings [132], not possible by natural means in a tunnel configuration unless done manually and periodically by appropriate maintenance of the infrastructure (best done on a weekly basis, also because the lamps will be covered with dirt and dust from the traffic (see, e.g., Figure 57), thereby diminishing the UV irradiance of the lamps—compare Figure 49b and Figure 50).

The principal goal of the European funded PICADA project (Photocatalytic Innovative Coverings Applications for Depollution Assessment) [134] was the development of a range of innovative materials that could easily be applied to various structures (buildings, tunnels, streets, roadside noise barriers, etc.) and that possessed despoiling and depolluting capabilities such as the TiO₂ semiconductor photocatalyst (in powder format) that had earlier shown excellent photocatalytic properties toward the oxidative/reductive destruction of organic contaminants in aqueous media and in air (VOCs), removal of toxic metals (Hg and Pb), recovery of precious metals (Ag, Au, Pt, Pd, and Rh) [144,145,146,147,148], and self-cleaning capabilities when deposited on glass and other supports (in colloidal format). The PICADA project involved a consortium (eight partners) that assembled both industry and research institutes from Greece, France, Italy, Denmark and Great Britain, and whose two main objectives were: (i) to develop and optimize industrial formulations of TiO₂ and application methodologies; and (ii) to establish local behavior models under different exposure conditions in realistic urban environments [134].

Various cement-based and ready-to-use products within this project involved nano-sized TiO₂ and suitable additives/binders with products to be developed with regard to desoiling and depolluting efficiencies, ease-of-use, durability, and cost effectiveness;

Table 8. Preparations and properties of the range of water-based products developed within the PICADA Project.

Products	Notation	Precursor Materials	Properties	Applications
Dry mix (water added on site)	B1	Sand; White ordinary Portland cement; lime; TiO2; mineral binder	10 mm thick decorative mortar	Façade coatings; Interior applications
	B2	Filler; White ordinary Portland cement; TiO2; mineral binder	1 mm mineral rendering; 100 μm thick paint	Thin decorative mineral façade coating; paint; Interior applications
Ready-to-use	C1	Siloxane binder; TiO2; CaCO3 filler	Translucent; resistant to photocatalytic effect of TiO2	Coating
	C2	Siloxane binder, TiO2	Opaque coating	Paints
	C3	Acrylic binder; TiO2	Opaque coating	Paints; Indoor applications
	C4	Silicate binder; TiO2	Opaque coating	Paints

gives a brief description of these products.

Laboratory tests of these products toward NO_x destruction were carried out in a 0.45-m glass chamber equipped with a UV lamp 50 cm above the eight petri dishes and a fan to circulate the inlet air/NO_x mixture (NO_x, 200 ppb); the materials displayed significant performance toward destruction of NO_x (especially NO) that disappeared completely at the end of the experiments. By comparison, tests toward the destruction of NO₂ were carried out in a 20 L glass chamber with one petri dish per experiment; UV lamp was located 25 cm above the dish, air/NO₂ mix (NO₂, 200 ppb) was recirculated within a closed circuit; nitrite and nitrate ions were detected. Materials again performed well toward destruction of NO₂ with an efficiency ca. 10 times greater than a control material without TiO₂ [134].

Subsequently, three of the products (the 10-mm thick B1 mortar, the 1-mm thick B2 mineral coating, and the translucent C1 coating) were coated on a 4-m² glass surface and tested in a much larger chamber (23 °C; relative humidity, 50%) equipped with a fan to keep the air/NO_x (NO_x, 200 ppb) circulating; lamp was located 150 cm above the samples. Results of the latter tests showed that in all cases ca. 80% of NO was destroyed. The TiO₂-treated B1 mortar was also tested in a pilot-scale three-canyon streets (dimensions: 18 m long, 2 m wide and 5 m high) near Paris; NO_x gas was emitted from an engine into the test and reference canyons by a perforated pipe. The B1 mortar reduced NO_x by 40–80% relative to the reference canyon without the B1 test material [134]; another report with more details is also available [149].

Gurol [150] examined the results of the PICADA project and concluded (at that time) that there were several unanswered questions that needed to be addressed in a laboratory environment with regard to reaction rates, mechanisms, reaction products, fate of reaction products, and types of pollutants; the effect of various variables that include TiO₂ particle size, type of TiO₂, percent loading of TiO₂ in the mixture, thickness of the mixture required (penetration distance of reactions), temperature, humidity, and concentration of pollutants; identification and quantification of reaction products under various indoor and outdoor conditions to establish that no hazardous chemicals were released from photocatalytic reactions; determination of the useful life of the TiO₂-containing material; and evaluation of the effectiveness of TiO₂ in a colored matrix as all buildings are not necessarily painted white. In addition, laboratory experiments ought to be conducted in a fully-controlled system that can operate under steady-state and continuous-flow conditions (as opposed to batch lab experiments) to be able to simulate realistic conditions. Furthermore, Gurol [150] recommended that a conceptual process model be developed to describe mass transport and reaction of various pollutants under various realistic scenarios for outdoor atmospheric conditions, together with evaluating the sensitivity of the model to all possible variables and atmospheric conditions. Whether these issues were considered and examined is typically not disclosed by industries, as they too often maintain that the results are proprietary.

Nonetheless, additional concerns with regard to outdoor applications remain to this day: (i) the long-term effectiveness of TiO₂-containing materials; (ii) the possible desorption of pollutants at night time; (iii) the decrease of the effectiveness of TiO₂ over time if adsorbed reaction products and chemicals are not washed off from the surface; and (iv) formation of ozone during the degradation of NO₂ as reported by Maggos et al. [151].

Field studies undertaken in 2005–2009 in the Netherlands by an international panel of air quality experts sought innovative solutions to improve air quality on and around motorways in densely populated bottlenecks (hot spots) along some of the Dutch motorways. To this end, two series of practical trials were conducted using four panels with different TiO₂ coatings fitted to an existing noise barrier along the A1 highway at Terschuur, and on a later erected TiO₂-coated porous concrete noise barrier along the A28 motorway at Putten (Figure 51); measurements of NO₂ and NO_x performed under various weather conditions at both locations showed very low conversion rates of the NO_x gases [152] that were attributed to the short contact time between air and barrier, to the relatively unfavorable meteorological conditions (wind direction and light intensity), to the high relative humidity, and to frequent low temperatures in the Netherlands. The experts concluded that improved air quality with TiO₂ coatings on concrete had not been demonstrated!

Road rehabilitation of Petersbergerstraße in Fulda (Hesse, Germany) was undertaken to investigate possible effects on the concentration of NO₂ emission by TiO₂-coated photocatalytic paving stones installed on sidewalks (ca. 800 m) on both sides of the street [153]. Measurements were carried out at two points (Points 1 and 2) across from each other and taken in June 2010 (Point 1) and in October 2010 (Point 2) at two different heights: 10 cm and 3 m above the road surface; for comparison, similar measurements were undertaken beyond the 800-m stretch as a reference. The average traffic volume on this street amounted to about 24,000–30,000 vehicles per day; proportion of light and heavy trucks and buses was 3.2% and 2.1% and 0.2%, whereas the car share was 88%, and motorcycles 6.5%. The whole TiO₂-coated surface was ca. 4500 m², a relatively small proportion compared to the remaining road surface of ca. 15,000 m². Relative to the reference section, the NO₂-reducing effectiveness appeared somewhat greater directly above the pavement surface (at 10 cm) than at the 3 m height; at Point 2 the effectiveness was, respectively, 17% and 9%, whereas at Point 1 the effectiveness was 3.5% at 10 cm above ground while the NO₂ level remained virtually unchanged at 3 m above. The large fluctuations in the individual values cast doubt as to whether the results were significant [153]. It must be noted that dirt, dust and oily layer on the sidewalks caused by the traffic and pedestrians had a non-insignificant impact on reducing the level of NO_x emitted. By contrast, Cristal Global (producer of CristalACTiV™; anatase TiO₂ nanoparticulate powder or colloidal sols) carried out a trial in 2006 in which TiO₂ was incorporated into paving slabs laid down at either side along the length of a street (covering ca. 1200 m²) in the Borough of Camden (London, UK); a chemiluminescence analysis of NO_x showed a reduction of ca. 20% (no other details were given) [154].

A field trial conducted at a train station in Manila (Philippines) by coating a 4100-m² exterior wall with TiO₂-based photocatalytic paint (Cristal Global) showed that about 26 g of NO_x per 100 m² of painted surface was removed; it was also claimed that each painted square meter could remove 80 g of NO_x per year [154,155]. In another trial that ran over a four-year period, a 135 m² wall in London was treated with a Cristal photocatalytic paint; the company claimed a reduction of 60% of NO and 20% of NO₂ in the vicinity of the wall (again short in details regarding the exact physical steps taken in the trials). In addition, a depolluting 12-month trial in an indoor car park in 2007 in Paris, in which two Cristal specialist paints were used with a very active photocatalyst within the matrix of the coating, revealed (analysis of monthly nitrate accumulation) reduction levels of NO_x between 53% and 99% depending on paint type and lighting levels [154]. However, another trial carried out, this time in a courtyard behind the Central St. Martin’s College of Art & Design (London, UK) by Cristal Global with one of its photocatalytic coatings, showed that, subsequent to chemiluminescence monitoring of NO, NO₂ and NO_x for nearly two years after the application of the photocatalytic coating, NO_x reductions amounted to 35–65% depending on time of year and local weather conditions [154].

With regard to processes that might take place on paint surfaces, Laufs and coworkers [156] examined the photocatalytic reactions of nitrogen oxides (NO_x = NO + NO₂) on commercial TiO₂-doped façade paints in a flow-tube photoreactor under simulated atmospheric conditions in a laboratory setting. Both NO and NO₂ were rapidly converted photocatalytically, albeit only on the photocatalytic paints but not on non-catalytic reference paints. Nitrous acid (HONO) was formed in the dark on all paints examined; however, HONO decomposed efficiently under irradiation only on photocatalytic samples, so that photocatalytic paint surfaces did not represent a daytime source of HONO, contrary to pure TiO₂ surfaces. Formation of adsorbed nitric acid/nitrate anion (HNO₃/NO₃⁻) occurred with near unity yield. The mechanism proposed by Laufs et al. for the photocatalytic reactions of NO, NO₂, and HONO carried out on photocatalytic paint surfaces in a laboratory setting is displayed in Figure 52 [156].

A full-scale outdoor field demonstration of air purifying pavement in Hengelo (The Netherlands) was carried out by Ballari and Brouwers [157] on the full width of a street surfaced with a concrete pavement containing C-doped TiO₂ (Kronos International; 4% w/w TiO₂ water suspension; 50 L of suspension covered a 750 m² surface; TiO₂ loading was 2.67 g m⁻²) sprayed over a length of 150 m; for comparison, another part of the street (ca. 100 m) was paved with normal paving blocks; outdoor monitoring was done during 26 days over a period of more than a year. Prior and during the field measurements, the used blocks were examined simultaneously in a laboratory setting (small flow-through reactor) to assess performance. The first coating applied May 2010 gave good results in the laboratory setting (7.7% of NO and 6.9% of NO_x under visible light illumination) and in the field. NO_x levels were sampled at different heights: 5 cm (near the active surface), 30 cm (car exhaust height) and 1.5 m (the breathing zone) to assess the extent of deNOxing as a function of distance from the active surface. Unfortunately, the TiO₂ photocatalytic coating vanished after the blocks were exposed outdoors for 2.5 months due to normal wear, to vehicular traffic, to weather, and/or else due to solid dust and dirt deposits on the surface. A second coating was subsequently applied to the surfaced road in September 2010; after being exposed for 1.5 months to the street environment, the photocatalytic performance returned to values of the first coating (laboratory testing). On average, the extent of NO_x converted (to nitrates) outdoors determined by the chemiluminescence technique was 19.2 ± 17.8% (daily readings) and 28.3 ± 20.0% (afternoon readings)—note the high fluctuations in the samplings, not typically reported by many researchers; under ideal weather conditions (high radiation; low relative humidity) the decrease of NO_x was 45% [157].

We have seen thus far that although NO_x levels can be reduced effectively with TiO₂-based photocatalytic surfaces in an indoor laboratory setting, significant variable results have been more the rule than the exception with regard to NO_x reduction levels in an outdoor urban setting. For instance, NO_x reduction levels of 40–80% [134,136,149], 26–66% [137], and ca. 19% [157] have been reported in various outdoor field trials; however, as we have seen in the field trial in Putten (The Netherlands) [152], the measured NO_x reductions were at or well below detection limits. Such variations are likely the result: (i) of limited contact between the pollutant NO_x and the photo-catalytic surface; (ii) of variable features of field sites; (iii) of prevailing local atmospheric conditions (wind velocity, wind direction, relative humidity, light intensity, etc.); (iv) of the time of measurements; and (v) of the nature and source of TiO₂ photocatalytic products and their associated reactivity upon photo-activation.

Attempts to reduce NO_x levels in an external “indoor” environment such as a road tunnel, as was the case with the Umberto I tunnel in Rome [123,138], were expected to bring about certain benefits: reduced health issues of pedestrians that use the tunnel, no need for a ventilation system, and decrease of the contribution of pollutants to the surrounding areas. To ascertain the benefits of photocatalysis toward NO_x reductions in such indoor infrastructures and as part of field studies within the PhotoPAQ program, Gallus and coworkers [14] undertook an extensive field study of the Leopold II tunnel in Brussels (2011–2013) where photocatalytic cementitious coating materials (first campaign; Italcementi’s TX-Active^® Skim Coat) were applied on the side walls and ceiling of a test section (73 m long) of one tube of the Leopold II tunnel (Figure 53) [158], followed by a monitoring campaign and later by a third campaign in which the same section was extended to 160 m and covered by a novel photocatalytically more active mortar (Italcementi’s TX-Active Skim Coat Boosted) [14]. Installed UV lighting (wavelength range, 315–420 nm) had an average irradiance on the active surfaces of 0.6 ± 0.3 W m⁻² (TX-Active; second campaign) and 1.6 ± 0.8 W m⁻² (TX-Active Boosted; third campaign).

Most interesting in the field trial of Gallus et al. [14] was the highly recommendable experimental approach used to assess the level of photocatalytic reduction of NO_x gases inside the tunnel based on the realization that changes in atmospheric compositions, traffic flow, and dilution (i.e., different wind speeds) occur inside the road tunnel. Consequently, the authors monitored NO_x concentrations and normalized them to the photocatalytically inert tracer CO₂ gas that was emitted by the vehicular traffic at the same time as the NO_x (and others) to yield NO_x/CO₂ concentration ratios; the instruments used throughout the campaigns had typical detection limits of 1–2 ppb, precisions of ~1% and accuracies of ~7% and ~10% for NO and NO₂, respectively; CO₂ levels were measured by a nondispersive infrared absorption technique: detection limit, 2 ppm; precision, 1%; accuracy, 7%.

The NO_x/CO₂ ratios so obtained from the slopes of plots of NO_x against CO₂ were independent of the absolute pollution level and thus of the emissions from the vehicular source and variable dilution inside the tunnel. Comparison of such slopes obtained in a control setting (dark conditions or otherwise under light illumination but without the TiO₂ photocatalyst) relative to a setting in which the slopes of NO_x versus CO₂ plots obtained under light illumination should show significant variations if NO_x were photocatalytically converted (to nitrates and others, such as HONO). Moreover, changes in the NO_x/CO₂ ratio—e.g., between upwind and downwind of the active section—could then be attributed to a photocatalytic remediation of the pollutants. The demonstration of NO_x levels at different heights within the tunnel were also relevant in their approach; in this regard, Figure 54 shows NO_x levels at Site 2 (under normal traffic conditions, only air at this site was in contact with the photocatalytic surfaces) were sampled at 1.1 m and 3.2 m above the street level and then plotted against the corresponding data at the tunnel ceiling (4.4 m above the street level) [14]. No significant gradients were observed at Site 2 under normal driving conditions with similar concentrations at the different heights above the street (note the slopes of the lines are nearly unity for data at the two heights).

Plotting the NO_x data from the downwind Site 2 of the third campaign against CO₂ data yielded a slope that gave a NO_x/CO₂ ratio of (3.09 ± 0.04) × 10⁻³ (Figure 55). More comprehensive results obtained under various conditions are summarized in

Table 9. Average NO_x/CO₂ ratios obtained from experimental results in the photocatalytic remediation of NO_x in the Leopold II tunnel in Brussels; standard errors represent 2σ **.

Average NOx/CO2 Ratios from Site 1 and Site 2 in the Second and Third Campaigns with Photocatalytically-Active Surfaces and with Lights ON
Campaign	Photocatalytic Material (TiO2)	Site 1—upwind (×10−3)	Site 2—downwind (×10−3)
Second; 09/2011	TX-Active	3.03 ± 0.06	3.14 ± 0.07
Third; 01/2013	TX-Active Boosted	3.18 ± 0.08	3.10 ± 0.05
Average NOx/CO2 Ratios for High and Low Wind Speeds with Lights ON in the Third Campaign
Wind speed		Site 1—upwind (×10−3)	Site 2—downwind (×10−3)
Low	<2 m s−1	3.24 ± 0.13	3.23 ± 0.12
High	>2 m s−1	2.93 ± 0.20	2.95 ± 0.10
Average NOx/CO2 Ratios at Downwind Site 2 with Lights ON or OFF in the Third Campaign
Data used		Lights OFF	Lights ON
All		3.09 ± 0.06	3.10 ± 0.05
Low wind speed	<2 m s−1	3.25 ± 0.10	3.23 ± 0.12
High wind speed	>2 m s−1	2.99 ± 0.11	2.95 ± 0.10

** Adapted from Gallus et al. [14]. Copyright Elsevier Ltd. (License No.: 4453120115447).

[14].

Comparison of all upwind (Site 1) and downwind (Site 2) data from the second campaign showed a minor formation of NO_x (70 m test section; TX-Active used), whereas a small reduction was inferred at Site 2 from the third campaign (160-m section; TX-Active Boosted used) in qualitative accord with laboratory experiments on sample plates exposed to tunnel air. In the latter experiments, the authors [14] observed photocatalytic formation of NO_x on dirty tunnel samples when the TX-Active material was used (second campaign), whereas only a very small reduction in NO_x was measured on sample plates exposed to tunnel air when TX-Active Boosted was used. An examination of the combined errors of the NO_x/CO₂ ratios together with the precision of the duplicate instruments used (±2% for NO_x) (see Table 9) led the authors to deduce that the observed differences in results from the second and third tunnel campaigns were insignificant.

To substantiate whether there was a potentially small photocatalytic reduction of NO_x from the upwind/downwind data from the third campaign, the authors [14] further examined the influence of reaction time on the photocatalytic NO_x remediation in the tunnel, in which a greater photocatalytic reduction was expected at lower wind speeds because of longer reaction times (residence times). However, no photocatalytic reduction of NO_x was observed on comparing upwind and downwind data for both sets of wind speeds (or reaction times), thereby precluding any (expected) increase in reduction of NO_x pollution at the lower wind speed. Comparing the NO_x/CO₂ ratios obtained with lights ON (active) and lights OFF (non-active) also showed no quantifiable remediation beyond experimental uncertainty, as also observed when only low wind speed data were used for which highest reduction was expected (Table 9). Accordingly, in view of the precision errors in the data analysis, the authors concluded that the extent of photocatalytic NO_x remediation in the 160 m more active tunnel section from the third campaign was at best ≤2% [14].

A theoretical model showed that, in accord with the 20% NO_x reduction observed in the 350 m long tunnel in Rome [123,138], a reduction upper limit of ≤20%, would have been possible in the 160 m test section of the Leopold II tunnel conditions as relative humidity (RH, 50%), wind speed (1 m s⁻¹) and UVA light irradiance (10 W m⁻²) were optimal, and deactivation of the tunnel’s photocatalytic surfaces was disregarded. Unfortunately, deactivation of these surfaces was not insignificant under the heavily polluted tunnel conditions as demonstrated by laboratory experiments, not to mention that UVA irradiances of 0.6 and 1.6 W m⁻² were far below the targeted value above 4 W m⁻², which was therefore a contributing factor in the deactivation phenomenon and further decreased the photocatalytic activity. Moreover, typical wind speed (ca. 3 m s⁻¹ during daytime) and the cold and humid (RH, 70–90%) conditions during the third campaign of January 2013 also caused a decrease of the activity of the photocatalytic material. Interestingly, another simple model calculation [14] that used uptake kinetics determined from laboratory experiments under the polluted tunnel conditions indicated an upper limit of only if such 0.4% for the photocatalytic NO_x remediation, in fairly good accord with the experimental tunnel results of <2%. Not to be neglected, photocatalytic degradation of NO led to significant formation of NO₂, particularly under humid conditions, thereby lowering the expected NO_x reduction even further.

Thus far, the description of some field trials carried out under a variety of experimental approaches and conditions that could not be controlled, as the trials were performed in an open air environment, have led to a large variation in results associated with the extent of TiO₂-based photocatalytic removal of NO_x gases. Accordingly, it is worth summarizing the results, which are collected in

Table 10. Percent removal of NO_x gases in field trials carried out with TiO₂-based photocatalytic protocols in an open outdoor environment and under a variety of conditions.

Field Trial	Gas	Extent of Photocatalytic Removal	Structures	Year	Ref.
1	NOx	37–82%	Model street canyon	2008	[136]
2	NOx	30–40%	Local street	2007	[137]
3	NOx	20–50%	Street	2012	[138]
4	NOx	45–57%	Street	2007	[137]
5	NOx	≤2%	Model street canyon, Tunnel	2015, 2015	[14,15]
6	NOx	> 20%	Tunnel	2012	[138]
7 (a)	NO	55%	Church external wall	2004	[139]
	NO2	32%	Church external wall
8	NOx	20–25%	Tunnel	2012	[138]
	NOx	51–64%	Tunnel
9	NOx	40–80%	Model street canyon	2006, 2008	[134,136,149]
10	NOx	Not-measurable	Highways’ noise barriers	2010, 2009	[152,160]
11	NO2	0–17%	Sidewalks/street	2012	[153]
12	NOx	20%	Street	2006	[154]
13	NO	60%	Building external wall	2006, 2012	[154,155]
	NOx	20%	Building external wall
14	NOx	53–99%	Car park	2006	[154]
15	NOx	35–65%	Building external wall	2006	[154]
16	NOx	1–37%	Street	2013	[157]
	NOx	8–48%	Street
17	NOx	26–66%	Real urban street canyon	2007	[137]
18	NOx	25–30%	Model street canyon	2010	[159]
19	NOx	40–80%	Model street canyon	2006, 2008	[136,149]
20	NOx	19%	Real urban street canyon	2013	[157]
21	NOx	Not-measurable	Street	2012	[153]

(a) Photocatalytic material (TX-Active) tested in a laboratory setting.

[14,15,134,136,137,138,139,149,152,153,154,155,156,157,159,160]. Clearly, the results are all over the place.

It has been reported time and again that photocatalysis, especially with TiO₂-based commercial products, could degrade environmental pollutants as shown, within the present context, by the removal of NO_x gases from the environment not only in a laboratory setting (Table 7), but to some extent also outdoors under environmental conditions (Table 10). Taking artificial model street canyons as examples for the removal of NO_x, we have seen that the reported extent of NO_x removal varies from 25–30% to 40–80% and 37–82%, while in a real urban street canyon values of 19% and 26–66% have been reported (Table 10). By contrast, similar TiO₂-based products have shown no effect in removing environmental NO_x gases in urban streets or on the highways’ noise barriers.

All the outdoors field trials expected the NO_x to be transformed into nitrates on the photocatalytic surfaces ultimately to be desorbed when raining. However, there are also reports that other intermediate species are likely to form also such as nitrous acid (HONO) [156,161,162,163,164], which is far more toxic than the NO_x pollutants, and not least is the potential for reNOxification and formation of ozone from the reaction of adsorbed nitrates with reducing agents (TiO₂ conduction band electrons) [164,165]. For these very reasons and as part of the PhotoPAQ investigative program, Gallus and coworkers [15] structured a two-step campaign to investigate the fate of NO_x gases outdoors in two artificial model street canyons in Petosino (Italy) (see Figure 56). In the first campaign, both model canyons had their side walls covered with a photocatalytically inactive fibrous cement, while in the second campaign one of the canyons had its side walls and ground surface covered with a photocatalytic cementitious coating material (Italcementi’s TX-Active Skim Coat Boosted), while the other was used as the reference canyon; the experimental approach used was otherwise similar to the approach used to examine the fate of NO_x in the Leopold II tunnel [14]. The results of their investigations are reproduced herein to demonstrate, what we consider the best approach, their highly recommended protocol and their actual results, not simply the authors’ conclusions [15].

Results from monitoring and analyzing the NO, NO₂ and NO_x gases (10-min averaged values) from the active canyon were plotted against the results obtained in the reference canyon (Figure 57), with the expectation that differences in the respective slopes (

Table 11. Slopes from the plots of daytime and nighttime results for NO, NO₂ and NO_x gases from the active canyon versus the reference canyon; a negative value implies a photocatalytic remediation of that pollutant **.

Gaseous Pollutant	Daytime Results	Nighttime Results	(Day—Night) × 100 (% ± 2σ)
NO	0.953 ± 0.016	0.988 ± 0.029	−3.5 ± 3.3
NO2	0.982 ± 0.013	0.966 ± 0.014	+1.6 ± 1.9
NOx	0.984 ± 0.011	0.970 ± 0.014	+1.4 ± 1.8

** Reproduced with permission from Gallus et al. [15]. Copyright 2015 by Springer-Verlag Berlin Heidelberg (License No.: 4453121427505).

) between daytime data and nighttime data would reflect the photocatalytic effect independent of any artificial differences. Perusal of the slopes that are summarized in Table 11 indicates that only for the conversion of NO was there a hint of a photocatalytic effect (−3.5 ± 3.3%). In addition, comparison of the daytime and nighttime results led the authors [15] to infer that less than 2% (that is, ≤2%) of the NO_x was converted photocatalytically to nitrates (

Table 12. Average concentrations of NO, NO₂, and NO_x in both canyons for daytime (06:00–20:30) and nighttime (20:30–06:00) and relative concentration differences; negative values imply a photocatalytic effect in the active canyon **.

		Concentrations (ppb ± 2σ)
		NO	NO2	NOx
All data	(1 − (Ref./Act.) × 100 [%]	2 ± 17	−0.7 ± 6.1	−0.4 ± 6.0
Daytime	(1 − (Ref./Act.) × 100 [%]	0.6 ± 11.5	0.2 ± 6.1	0.3 ± 6.0
Nighttime	(1 − (Ref./Act.) × 100 [%]	23 ± 150	−2.4 ± 5.9	−2.0 ± 5.9

** Adapted with permission from Gallus et al. [15]. Copyright 2015 by Springer-Verlag Berlin Heidelberg (License No.: 4453121427505).

).

The field trial carried out in the Leopold II tunnel in Brussels found that the photocatalytic active surface was deactivated [14]. Accordingly, laboratory tests were undertaken to verify whether deactivation might also have occurred in the canyon trial campaign by taking samples of the active surface in the canyon (before and after) and using a flow-through reactor to examine their photocatalytic behavior toward degradation of the gas pollutants. Results showed that no deactivation of the canyon surfaces had occurred. Consequently, the authors [15] surmised that this was due to several evident differences in the model canyon campaign versus those from the Leopold II tunnel; that is,

• Much lower pollution levels exist in the canyon than in the tunnel; in the latter case, the greater pollutant level (very dirty ambient conditions–dirt/dust/grime/air pollutants) blocked the active sites (see Figure 58).
• Much higher UV irradiance was available in the canyon (sunlight UV in the canyon versus an artificial UV light system in the tunnel) causing the pollutants to degrade faster.
• In the tunnel case, the cementitious surfaces were allowed to cure in the dark for several days, whereas photocatalytic and non-photocatalytic surfaces in the canyons were freshly exposed to sunlight UV radiation.

Even though they expected to confirm NO_x removal levels of ca. 20%, if not greater (see Table 10), as reported in earlier field trials in open atmospheric environments, the findings that there was no significant photocatalytic remediation of NO_x (and VOCs, O₃, and particulate matter)—removal level of ca. 2% or less—was somewhat enigmatic to say the least [15], albeit consistent with non-measurable NO_x reduction levels found in highways (The Netherlands) and in a real urban street (Fulda, Germany). Such minute NO_x reductions were likely due to low surface-to-volume ratios (S_active/V) of the open structures; note that the S_active/V ratio is a very significant factor that reflects the available active surface versus the space volume occupied by gaseous pollutants, i.e., the limiting heterogeneous uptake of a species onto a solid surface [166,167], as is also the associated deposition velocity. Evidently, other factors must have played a role, and/or some differences in experimental approaches might aid in explaining such very low levels of NO_x reductions, namely:

• Although the overall geometry of the PhotoPAQ canyon model was otherwise similar to earlier model street canyons, the dimensions were different with respect to smaller model sites (e.g., 5 m wide canyon versus the earlier 2.4-m wide canyons that led to unrealistically higher S_active/V ratios and as such to higher NO_x reduction levels of 20–80% (Table 10) versus ≤2%. Note that the large reduction levels of NO_x on photocatalytic TiO₂ surfaces experienced in laboratory settings (Table 7) are also explained by this surface-to-volume ratio factor which is considerably greater: typically, 167 m⁻¹ for a small flow-through reactor in a laboratory setting versus 0.6 m⁻¹ for the model street canyon in the open environment in the industrial zone of Petosino (Italy). If one were to scale down the earlier results [134,136,149,166,167] to real urban street canyon conditions, it would result in only approximately 5% NO_x reduction.
• Previous field trials considered results of monitoring the fate of NO_x pollutants during daytime hours only, contrary to other field trials where the whole diurnal data were considered. NO_x levels during nighttime were often comparable to those during daytime (higher emissions during the day were often compensated by stronger convective vertical dilution) so that the above estimate of ~5% based on daytime data was further reduced to ca. 3%, comparable to expectations typical of a main urban street [15].
• The distance at which the NO_x gases were sampled from the photoactive surface is another important factor. Gallus et al. [15] recommended sampling NO_x at a distance of 3 m above the photoactive surface for urban network stations, contrary to earlier field trials where sampling of NO_x was done but a few centimeters from the active surface (5 cm to 1.5 m [148]; 30 cm to 1 m [137]; 50 cm [166,167])—indeed, since conversion of NO_x gases to nitrates requires the NO_x to be adsorbed onto the photocatalytically active surface, it is obvious that close to the surface is where the greatest change in NO_x concentration will be felt—this means that, as you stroll on a photocatalytic street surface, your feet will experience a healthier environment than your head.
• High reductions of NO_x levels in earlier field trials [134,137,149,157,166,167] likely reflected some differences between active sites and reference sites in terms of the quantity of NO_x artificially injected into the active and reference canyons [134,136,149,166,167], whereas the field trial in the Petosino model canyon the NO_x gases were those present in the homogeneous industrial environment [15].
• Where the active and reference model street canyons were distant from each other [137,157] may also have resulted in non-insignificant differences in the quantity of NO_x artificially injected and in pollution dispersion (e.g., different wind speeds and different wind direction), thus leading to non-insignificant uncertainties in the results.

With regard to the latter issue, model studies have re-evaluated the daytime results from the experimental street canyon in Bergamo and determined that the upper limit of photocatalytic NO_x remediation was more like 4–14% [160], rather than the claimed 26–66% [137], because of: (a) strong differences in the vehicle NO_x emissions at the active and reference sites with much higher vehicle fleet density at the reference site; and (b) different dispersion conditions (geometry of sites, micrometeorology). However, Gallus et al. [15] contended that, if in the model studies one were to assume a reasonable wind speed of 1 m s⁻¹ and a realistic photocatalytic deposition velocity of NO_x on the active surface of 0.3 cm s⁻¹, the upper limit of 4% [160] would be reduced further to <2% considering diurnal averages and transport limitations and more in line with real environmental results [152,153,168]. Clearly, the above factors and discussion call attention to the necessity of some care into how the data are treated and how the data are interpreted.

Several TiO₂-based photocatalytic materials were also tested in 2015 by Pujadas and coworkers [169] of CIEMAT (Madrid, Spain) within the framework of the 2013–2018 LIFE MINO_x-STREET European project. Tests were first carried out in a laboratory setting to choose the most photoactive TiO₂-based material (fresh sample showed 45% reduction of NO in a flow-type reactor, dropping to 20% for a used street sample caused by traffic and ageing over a month) so designed as to be used on bituminous mixtures with which to examine the effect of reducing NO_x levels in the two-way Boulevard Paseo de la Chopera, a real urban setting in the Municipality of Alcobendas (Madrid)—conditions were: ground level wind speed, 2 m s⁻¹; relative humidity, <65%; solar radiation, >400 W m⁻²; TiO₂-based coated active area, 1000 m²; length of section, ca. 60 m. The experimental approach was similar to that used in the PhotoPAQ project in that the daytime and nighttime NO_x collected data in the median strip at two active sites (Site 2 and Site 3) and at the inactive site (Site 4) were plotted against NO_x data collected at the inactive Site 1; the plots yielded the slopes reported in

Table 13. Slopes from correlation plots of NO data (ppb ± 1σ) from three sampling sites (two active, one inactive) against inactive Site 1 for NO concentrations before and after implementation of the photocatalytic coating.

	Diurnal		Nocturnal
Sampling Site	Before Applying TiO2-Based Coating	After Applying TiO2-Based Coating	Before Applying TiO2-Based Coating	After Applying TiO2-Based Coating
2 (active)	1.05 ± 0.08	1.07 ± 0.08	1.014 ± 0.005	0.983 ± 0.004
3 (active)	1.41 ± 0.17	1.03 ± 0.06	0.977 ± 0.008	0.915 ± 0.004
4 (inactive)	0.89 ± 0.06	0.90 ± 0.07	0.985 ± 0.005	0.946 ± 0.025

[169]. The similarities between the data collected in daytime and at nighttime during the most optimal measurement periods before and after application of the photocatalytic coating led the researchers to deduce that there was no photocatalytic reduction of NO levels, in line with observations made by others in a real urban setting (see above).

A real scale examination of NO_x depollution by TiO₂-based photocatalytic sidewalk pavement and façade was performed by Pujadas and coworkers of CIEMAT [170] in an urban environment that resembled a model street canyon, albeit narrower, in which they also found no significant NO_x reduction during their experiments in the sidewalk model in open air, even though in a laboratory setting the extent of NO_x removal by the photocatalytic material was 65%. Nonetheless, a small ambient NO_x reduction effect was observed on the photocatalytic façade (mimicking a building brick wall), although for only a very short time and very close to the active surface (e.g., <10 cm), and then only under very specific ambient and meteorological conditions.

At a presentation of their research at a conference in Barcelona (Spain) that was performed within the MINOx-STREET project, Palacios and coworkers [171] described results on urban street experiments, as well as experiments with model sidewalks and model building façades. Two principal conclusions are worth noting from that conference:

• A great difficulty was experienced in urban scenarios to establish a possible cause-effect relationship between any observed ambient NO_x reductions in the presence of photocatalytic surfaces. In road bituminous pavement (the Alcobendas experiment) and sidewalk scenarios, the NO_x removal effect of photocatalytic materials had not been unequivocally demonstrated experimentally; however, in the case of the façade scenario that effect was documented, albeit the effect was seen only at distances very close to the wall surface.
• Physicochemical characteristics of heterogeneous photocatalysis, such as low quantum yield for absorption of solar radiation by TiO₂, heterogeneous molecular processes, and the high dependence on ambient conditions helped to explain the weak macroscopic effects observed in open urban air. Consequently, a low NO_x concentration reduction was only attained very close to photoactive surfaces with poor global incidence on ambient air.
• The CIEMAT group [171] also investigated the NO_x depolluting effect of TiO₂-based photocatalytic materials in a medium-scale tunnel reactor under semi-controlled conditions using 200 ppbv of NO and compared the results with those from a real-scale outdoor tunnel (street in Alcobendas) and from a laboratory-scale reactor—

  Table 14. Parameters and results of NO removal in a medium-scale tunnel reactor (UVA irradiance, >40 W m⁻²; relative humidity, <30%; dimensions, 0.4 × 0.4 × 10 m; photoactive surface, 0.4 × 10 m) compared to a real-scale tunnel and a small laboratory flow-through reactor.

  Parameters	Outdoor Real-Scale Tunnel Reactor	Outdoor Medium-Scale Tunnel Reactor	Laboratory Flow-through Reactor
  Deposition velocity (m s−1)	5.25 × 10−3	2.05 × 10−3	5.25 × 10−3
  Length (m)	60	100	0.1
  Air velocity (m s−1)	1	0.33	0.20
  Photoactive surface (m2)	1000	4	0.005
  Sactive/V (m−1)	0.067	2.5	200
  Residence time (s−1)	60	30.3	0.5
  First order rate coefficient (s−1)	3.4 × 10−4	5.07 × 10−3	1.1
  Average yield of NO removed (%)	~2	15 ± 4	41

  summarizes the parameters and the extent of NO removed in all three cases [171].

6. Concluding Remarks

One of the objectives of this review article is to provide an introduction to the basic approaches and terminology inherently and commonly used in heterogeneous photocatalysis. Relevant characteristic features of solid semiconductor/insulator photocatalysts are examined, together with some initial and subsequent events that follow the absorption of photons by these solids. The relationship(s) between traditional heterogeneous catalysis, on the one hand, and photochemistry and molecular spectroscopy, on the other hand, with heterogeneous photocatalysis have been made. The closer interconnection of photocatalysis to photochemistry is emphasized. Where similar symbols are used to mean different things, they are nonetheless fully described in the context used. The other objective is to examine closely the various attempts made to apply the photocatalytic technology—albeit briefly with regard to the published literature on environmental remediation of contaminated ecosystems—through incorporation of the extensively investigated TiO₂ photocatalyst into various cementitious substrates (concrete, mortar, plaster, paints, etc.) applied to various infrastructures (highways, urban streets, building external walls and road tunnels, among others) and through the use of various coating materials.

With regard to the latter aspect, we have witnessed the external conditions of the Dives in Misericordia Church and the Umberto I tunnel in Rome (Italy) over a decade later from when the photoactive coatings were first applied to these structures. Evidently, deactivation of TiO₂-based photoactive surfaces may become an issue over time so that, as Gallus et al. [15] aptly noted, experiments should be carried out at the field site of interest in small scale experiments exposed to the prevailing atmospheric conditions of the site—and not in a laboratory setting miles away from the application—before giving any consideration to apply the photocatalytic technology to surfaces of large infrastructures. Only when these surfaces display high activity towards NO_x (NO and/or NO₂) removal, in particular, and degradation of VOC pollutants, in general, and no strong deactivation of the photoactive surface occurs under the prevailing specific field conditions, should photocatalysis be advocated as a possible technology to reduce NO_x and VOC levels in open urban environments.

We have also seen that the photocatalytic technology has demonstrated significant removal of NO_x pollutants in a laboratory setting, whereas the overall results from real-scale application in an open air environment have been, to say the least, highly variable, but quite disappointing at best even under appropriate approaches in carrying out the experiments. Such deviations from a laboratory setting to a real outdoor environmental setting is due to several factors that advantage the laboratory results: (1) a small flow-through photoreactor was used in the laboratory tests against a vast open environment reactor system; (2) the photocatalytic surface used in the laboratory was a clean surface against an outdoor surface that in relatively short time is ultimately covered with dust, filth etc., thereby blocking the surface active sites where the reaction between the NO_x and the oxidizing entities at these sites takes place—while rain might wash off the nitrates, the dust, and the filth from the open outdoor, that is not the case where the indoor walls of tunnels are concerned; (3) the rate of flow of the NO_x gases could be controlled in the laboratory tests, unlike the open environment where no control of wind speed and/or its direction are possible; (4) the UV/Visible radiance could be controlled in the laboratory versus lack of control of the sunlight impinging on the photocatalytic surfaces; (5) unlike the open environment, in laboratory tests, the flow rate of the NO_x gases in the photoreactor could be controlled which, therefore, affected the deposition velocity of the NO_x onto the photocatalytic surface (i.e., a competition between horizontal flow versus vertical flow) as well as the residence time spent by the NO_x gases within the reactor; (6) the S_active/volume ratio in the laboratory reactor was far more advantageous than it could be in the open environment; (7) the relative humidity within the laboratory reactor could also be controlled, unlike the open environmental reactor; and (8) laboratory tests were carried out for a relatively short period (minutes to a few hours) within which the photocatalytic surface retained its integrity versus the open environment where the wished-for expectations were that the photocatalytic surfaces should remain active for months if not years and beyond.

Nonetheless, despite the disadvantages experienced in the open environment, there is no reason to throw away the baby with the bath water. It suffices to carry out cost-efficiency analyses (as they may be geographically different) and compare them with other technologies available out there—for example, selective catalytic reduction (SCR) and selective non-catalytic reduction (SNCR) together with natural means (e.g., trees on road sides). Photocatalysis may yet be found more attractive on a cost-efficiency basis compared to other, perhaps more expensive methods if extra costs of photocatalytic surfaces associated with application to normal urban surfaces (roads, paints, roof tiles, etc.) were minimized by the industrial sector in the future, and if photocatalytic materials were applied when urban infrastructures were being renovated [15].

It cannot be overemphasized that the effective use of photocatalysts in cementitious substrates in the deNOxification of the environment requires serious considerations of some fundamental physical and chemical notions prior to any attempt at implementation of the TiO₂-based photo-catalytic methodology, and as appropriately emphasized by Macphee and Folli [129], the following concerns are worth noting and sharing:

• Photocatalysis is a surface phenomenon that is impacted by the chemistry of the immediate environment, so that the cementitious photocatalytic surface must be engineered to maximize light absorption by the photocatalytic surface and for the reactants to access the surface.
• Redox potentials of pollutants considered for elimination must match the semiconductor band edge positions (i.e., flatband potentials) of the conduction and valence bands for a successful deNOxification as the band edges are pH-dependent, especially for a metal-oxide photo-catalyst.
• More fundamental research is needed toward optimizing photonic efficiencies of visible-light activation of photocatalysts if anion- and/or cation-doped photocatalysts were used, as geographical locations may be limiting the usage of photocatalytic concrete if conventional pristine TiO₂-based surfaces were used.
• There is a need to maximize both the dispersion of photocatalytic particles in the cementitious substrate and the photoactive surface area toward target applications and agglomeration of particles in a porous structure should be minimized as it could block access of larger pollutant molecules to the internal photoactive surface.
• Oxidative removal of NO_x pollutants by their conversion to nitrates occurs only if they are adsorbed on the photocatalytic surface; recall that the adsorption mode of molecules on the photoactive surface is highly pH-dependent (surface charge) and could thus affect the oxidative efficiency.
• The deNO_x index that combines photonic efficiency and selectivity into one environmental impact parameter for a given photocatalytic material should be considered whenever screening candidate photocatalytic materials; for instance, if this index for a photoactive surface were negative shows that the oxidative removal of NO favors formation of the more toxic product NO₂ rather than the NO₃⁻ species, as demonstrated for a conventional unselective TiO₂ (e.g., bare P-25 titania) that exhibited a large negative deNO_x index [129], a sign that the catalytic process was ineffective. Consequently, both photocatalyst selectivity and photo-catalyst activity must be assessed to reduce emission of harmful by-products.
• On the more fundamental side, the thermodynamics (energetics) are the most valuable tool in understanding and designing selective photocatalytic processes. Recall that, after the rapid e⁻/h⁺ pair recombination, the next fastest process is charge carrier trapping. This calls attentions that not only is the semiconductor band edge positions relevant, but also charge transfer from traps to adsorbate molecules need to be fully appreciated in interpreting photocatalytic processes as also the engineering and design of band structures of single semiconductors or semiconductor composites.

Additionally, using a tunnel as an example of an infrastructure in need of eliminating NO_x agents, Gallus et al. [14] also expressed the notion that a reasonable photocatalytic remediation of NO_x in a road tunnel can only be obtained under suitable/special experimental conditions, as photocatalytic remediation will not be effective where the photoactive materials are strongly deactivated under highly polluted conditions. Accordingly, photocatalytic materials should always be tested in small-scale bed-flow reactors on site using: (1) the UVA irradiance equivalent to that of the tunnel; and (2) the NO_x present in ambient air sucked from the tunnel as the reactant.

Only if NO_x were efficiently decomposed under these conditions, would larger scale applications in a road tunnel be worth considering. Regardless of the possible deactivation of photo-catalytic materials under high pollution conditions, model calculations have shown that applications of such materials are unlikely to result in significant remediation of NO_x to nitrates at low UVA irradiances and where high wind speeds and/or high relative humidity prevail. With regard to NO_x conversions in road tunnels, this calls further attention to the need for a careful characterization of tunnel conditions, for a quantification of possible deactivation of the photocatalytic materials, and for performing simple upper limit model calculations of expected NO_x reductions, together with a cost–benefit analysis well before considering application of photocatalytic materials and installation of costly UV lighting systems. These same considerations also apply to pollutant remediation in other infrastructures.

7. Recommendations

In practical applications of photocatalysis, conventional TiO₂-based photocatalytic surfaces have been used to oxidize NO_x to nitrate species; the latter species do not desorb spontaneously and consequently deactivate or block the surface-active centers of the photocatalyst from carrying out the next cycles. To avoid such deactivation, the nitrates (or nitric acid) should typically be washed away by rain [172]; however, the nitric acid is corrosive and could pollute the soil when its concentration at the site becomes too high. A promising way to resolve this problem, which the users of applied photocatalysis have failed to consider but known to occur in a laboratory setting since the first report by Courbon et al. [92] in 1984, would be to change the selectivity of the photocatalytic reaction so that the NO_x gases are converted back to N₂ and O₂ by some photoreduction pathway as reported some time ago by Anpo and coworkers [94,95], who used Cu⁺ ions in SiO₂ or in zeolite to effect the photocatalytic reduction of NO_x. No deactivation of the active sites would occur for this photoreduction reaction since nitrogen and oxygen readily desorb from the surface [173]. The selectivity toward the photoreduction of NO could be improved greatly by reducing the hexacoordinated Ti⁴⁺ species (TiO₆ octahedra) to tetracoordinated Ti⁴⁺ species (TiO₄ tetrahedra) [174], as successfully achieved by depositing isolated TiO₄ clusters inside cavities of zeolite-Y using ion beam implantation [174,175].

In a more recent article, Wu and van de Krol [176] proposed a novel strategy to change the photocatalytic selectivity of TiO₂ by creating a large and stable concentration of oxygen vacancies in TiO₂ nanoparticles through thermal reduction in a reducing atmosphere; these oxygen vacancies were stabilized by doping the TiO₂ nanoparticles with an electron acceptor-type dopant such as Fe³⁺ which also greatly enhanced the activity of the photoreduction process. The authors [176] further showed that with this strategy NO was indeed photoreduced to N₂ and O₂ and that photooxidation of NO was largely suppressed. Moreover, photoreducing Fe³⁺ to Fe²⁺ provided a recombination pathway that suppressed nearly quantitatively the formation of NO₂ and consequently enhanced the selectivity of the reaction for N₂ formation [176]. The authors also alluded to formation of N₂ and O₂ via two different routes. One route would see a small amount of tetrahedrally coordinated Ti formed in the Fe-doped TiO₂ samples, which Anpo et al. [175,177] claimed as the active site for the catalytic decomposition of NO to N₂ and O₂ at Ti-modified zeolites. As most Ti^IV ions at the TiO₂ surface are fivefold-coordinated, a single oxygen vacancy created at or near the surface could lead to a fourfold-coordinated Ti⁴⁺ center; however, this would require a strong reduction of the Ti–O bond length that would be possible only at very high oxygen vacancy concentrations, which the authors [177] deemed an unlikely pathway and proposed the other route that implicated oxygen vacancies acting as the catalytic centers through the capture of the oxygen side of NO as illustrated in Figure 59 and summarized in Reactions (101)–(104); the associated experimental data are also displayed in Figure 59.

V_o⁺⁺_(surf) + 2 e⁻ + NO_(g) → O_surf − N

(101)

2 O_surf − N → 2 O_surf + N_2(g)

(102)

2 O_surf → 2 V_o^x + O_2(g)

(103)

V_o^x + 2h⁺ → V_o⁺⁺

(104)

Summing Reactions (101)–(104) yields the overall Reaction (105).

2 V_o⁺⁺
2 NO_(g) + 4 hν → N_2(g) + O_2(g)

(105)

Although the conversion efficiency was somewhat modest (ca. 4.5% after 1050 min correspond-ing to a TON of ∼2 NO molecules per O vacancy site), the Fe-doped TiO₂ photocatalyst showed no signs of deactivation as the NO conversion centers were not blocked by nitrate species [168], contrary to standard deNO_x TiO₂-based photocatalysts that have to be washed away periodically.

In a most recent article, Cao and coworkers [178] investigated the adsorption of NO and the consequent reactions on differently treated rutile TiO₂(110) surfaces using polarization/azimuth-resolved infrared reflection absorption spectroscopy. Apparently, surface defects (e.g., oxygen vacancies, V_o) and reconstructions on TiO₂(110) had a strong effect on the reaction pathways of NO → N₂O conversion (N₂O is laughing gas). The pathway proposed involved a defect-free oxidized TiO₂(110) surface in which two NO molecules are adsorbed on adjacent surface-pentacoordinated Ti (Ti_5c) sites first, which then couple to form a cis-(NO)₂/Ti&Ti dimer through the N−N bond of the dimer, and then are converted to N₂O species (or perhaps even to N₂ gas) [178].

Clearly, much fundamental research in TiO₂-based photocatalysis needs to be undertaken in the optic toward applications to environmental deNOxification, with special attention and efforts directed at titania doped with Fe, Cr, Co and Ni dopants that may yet prove interesting [176].

As a case in point, a recent article by Kuznetsov and coworkers [70] examined possible additional specific channels of photoactivation of solid semiconductors with regard to thermo-/photo-stimulated bleaching of photoinduced Ti³⁺ color centers in visible-light-active (VLA) photo-chromic rutile TiO₂, which an optical emission spectroscopic analysis had shown to contain 99.4 at.% Ti and 0.2 at.% Al as the principal impurity, together with 0.09 at.% Fe, 0.05 at.% Sn, 0.04 at.% Nb, and 0.03 at.% Cr as minor impurities. Considering that the prime photophysical process of photostimulated bleaching of Ti³⁺ color centers is absorption of light quanta by the Ti³⁺ centers, the authors [70] found that no selectivity of photostimulated bleaching of a certain type of Ti³⁺ centers could be ascertained, and that photogenerated holes captured at a set of traps were also participants in the photostimulated bleaching of these color centers. Based on current findings and earlier results, the authors hypothesized that the heat released during nonradiative electron transitions, following the prime photophysical processes of excitation and ionization of Ti³⁺ centers, dissipates in the nearest neighborhood of the Ti³⁺ centers and that localized nonequilibrated excitation of the phonon subsystem leads to thermal detrapping of the photoholes with different depths up to 1 eV. Subsequent recombination of free holes with trapped electrons from Ti³⁺ centers leads to the observable photostimulated bleaching of the color centers [70]. Based on experimental evidence, the authors further argued that following absorption of vis–NIR light by the color centers, the subsequent release of thermal energy accompanying nonradiative electron transitions provides an additional specific channel to photoactivate the VLA rutile TiO₂, in particular, and possibly other photocolorable metal-oxide semiconductors as well.

Following their interest of the photophysics of color centers in VLA rutile titania ceramics and titania powder resulting from the photoformation and separation of charge carriers, Kuznetsov et al. [179] noted that the action spectrum of the photoformation of Ti³⁺ centers at very low temperatures (90 K) accorded fully with the absorption spectra of intrinsic defects that consisted of a set of individual absorption bands that they attributed to several different Ti³⁺ centers. Analysis of the dependencies of the photoformation of separate centers on the wavelength of illumination and light exposure, which provided extraction of specific Ti³⁺ centers, led the authors to identify Ti³⁺-based centers with excessive negative charge that formed at significantly high concentration upon maximal exposure of the titania specimens to Vis-light illumination: (2Ti³⁺ + V_o²⁺) ←→ (Ti^δ+ + V_o²⁺) with 3 > δ > 2. They also showed from thermoprogrammed annealing (TPA) that the spectra of Ti³⁺ color centers in the range 90–500 K consisted of a set of first-order peaks corresponding to traps, whose depths ranged from ~0.2 eV (peak at 130 K in the powder specimen) to 1.06 eV (peak at 455 K in the ceramics specimen). The highest rate of recombination of holes released to the valence band with Ti³⁺ centers—an event attributed to Ti^δ+ centers—provided TPA spectra that clearly manifested the existence of shallow traps. In addition, mass spectrometric experiments on the photoadsorption of molecular oxygen and photodesorption of photoadsorbed oxygen from the surface of powdered VLA titania specimens provided further evidence of the photoformation of electrons and holes in VLA TiO₂ under Vis-light illumination, and allowed the authors [179] to determine the kinetics of photo- desorption of O₂ under orange light illumination subsequent to photoadsorption of O₂ stimulated by blue light excitation. Those experiments provided further proof of the occurrence of another specific channel toward the photoactivation of VLA TiO₂ via photoexcitation of photoinduced Ti³⁺ color centers.

It is important to recognize that Ti³⁺-based centers (i.e., Ti^δ+ centers) appeared after many other Ti³⁺ centers had already been formed. In other words, such Ti³⁺-based centers appeared at high density of Ti³⁺ centers (see below). Accordingly, specific properties of Ti³⁺-related centers responsible for the (extrinsic) absorption bands at 1.56 eV and 1.26 eV were postulated to account for the excess negative charge characteristic of such Ti³⁺-based centers. In line with the work of Déak and coworkers [180], the two adjacent Ti³⁺ centers located near a single oxygen vacancy forming a (2Ti³⁺ + V_o²⁺) complex (Figure 60a) were taken by Kuznetsov et al. [179] as extra charged Ti³⁺ centers when compared to isolated Ti³⁺ centers (Figure 60b). The two Ti³⁺ species in the (2Ti³⁺ + V_o²⁺) complex can, in principle, disproportionate to (Ti²⁺ + Ti⁴⁺ + V_o²⁺) so that, in accord with the more generally accepted view, these extra-negatively charged Ti³⁺ centers are best referred to as Ti^δ+ centers for which 3 > δ > 2. Clearly, the appearance of such Ti^δ+ centers, whether photogenerated or resulting from the removal of the structural oxygen during a reduction event, is of lower probability because of the electrostatic repulsion of the two trapped electrons and the well-known instability of such centers to oxidation. Consequently, the formation and increase in the concentration of such photoinduced Ti^δ+ centers appears to occur only at high density of photogenerated Ti³⁺ centers that ensue upon prolonged exposure to Vis-light illumination in the later stages of photocoloration (i.e., formation of Ti³⁺ color centers).

Germane to the above, the work of Déak and coworkers [180] showed that the first case scenario is that two self-trapped electrons in the (V_o²⁺ + 2e) complexes are located at two equivalent first neighbors of the oxygen vacancy (extra-negatively charged Ti³⁺ → Ti^δ+ centers; Figure 60a), while in the second scenario both electrons are more remote from the V_o²⁺ vacancy and are not in the same plane as the vacancy (isolated Ti³⁺ centers; Figure 60b); the energies of the vertical transitions of these self-trapped electrons to the conduction band are ca. 1.1 eV. Following this reasoning, a question arose as to why the growth of the number of such Ti^δ+ centers was observed only under Vis-light illumination. This led Kuznetsov and coworkers [70] to focus attention on the differences in the spatial photoexcitation events that occur in the microparticle when illuminated in the UV and Vis spectral regions. Such differences had not heretofore been considered in the literature; their views of the events that occur under UV and Visible light illumination are summarized in Figure 61.

When hν > E_g, light quanta are absorbed spontaneously in solids in an arbitrary manner, each time producing e-h pairs at new spatial sites (black stars in Figure 61a) for which charge carrier transport and localization in the microparticle are determined by the distribution of charge carrier traps (Processes 2 and 3 in Figure 61a). At moderate UV-light irradiances, the authors [70] supposed that since every subsequent photoformation of charge carriers and trapping event occur in other spatial sites, a significant density of Ti³⁺ centers would not be reached. However, photoformation of electrons and holes can also be achieved on illumination in the Vis region at hν ≤ E_g when light quanta are absorbed by the native point defects (i.e., F (or F⁺) centers); the latter are limited in number and are located at definite sites in the microparticle (Figure 61b). Moreover, because photoexcitation of F (or F⁺) centers can produce charge carriers followed by their subsequent decay to their initial electronic states, as proposed in earlier studies [68,69], repetitive absorption of light quanta and photogeneration of electrons and holes occurs each time at the same spatial sites (F or F⁺ centers) in the microparticle (blue stars in Figure 61b). Such considerations then lead to the reasonable inference that transport of carriers and occupation of traps (processes 2 and 3 in Figure 61b) start repetitively at the same sites in the microparticle. In that case, filling of the nearest neighbor F (or F⁺) center traps facilitates the attainment of a high density of Ti³⁺ centers and the consequent formation of the Ti^δ+ centers [70].

The above notwithstanding regarding the TiO₂-based technology, people intending to deNOxify the environment must first come to appreciate and understand the rich chemistry of nitrogen oxides, in general, and NO and NO₂, in particular, in a homogeneous phase and in hetero- geneous media.

For instance, NO dimerizes to N₂O₂ upon condensing to a liquid, although the association is weak and reversible [181]. In addition, to the extent that the enthalpy of formation of NO is endothermic, NO can easily undergo disproportionation back to its constituent elements N₂ and O₂ as might occur in catalytic converters—for example, Reaction (106) occurs over the zeolite Cu²⁺-ZSM-5 [182].

2 NO → N₂ + O₂

(106)

Nitric oxide is also thermodynamically unstable at 25 °C and 1 atm; under pressure, it decomposes readily in the temperature range 30–50 °C to yield NO₂ and N₂O (Reaction (107)) and may react either as NO₂ or as N₂O₃ [183].

3 NO → N₂O + NO₂

(107)

When exposed to atmospheric oxygen, nitric oxide converts instantly to NO₂ (Reaction (108)) [181], which likely occurs via the intermediate ON–O–O–NO.

2 NO + O₂ → 2 NO₂

(108)

In water, NO reacts with oxygen and water to form nitrous acid HONO (Reaction (109)).

4 NO + O₂ + 2 H₂O → 4 HONO

(109)

Since both NO₂ and NO are radical species, they combine to form the intensely blue dinitrogen trioxide N₂O₃ (Reaction (110)) [184].

NO + NO₂ ⇆ ON − NO₂

(110)

Both the brown gas nitrogen dioxide, NO₂, and the colorless gas dinitrogen tetroxide, N₂O₄, exist in a strongly temperature-dependent equilibrium (Reaction (111)) for which ΔH = −57.23 kJ mol⁻¹, with NO₂ being favored at higher temperatures, while N₂O₄ predominates at lower temperatures.

2 NO₂ ⇆ N₂O₄

(111)

Because of the relatively weak N–O bond in NO₂, nitrogen dioxide is a relatively good oxidizing agent in aqueous media (Reaction (112); nearly comparable to Br₂ gas), which makes the mixed oxides NO₂ and N₂O₄—also known as nitrous fumes—react vigorously if not explosively with several compounds, particularly with hydrocarbons via hydrogen abstraction as a first step (Reaction (113)) [181].

N₂O₄ + 2 H⁺ + 2 e⁻ → 2 HONO    E⁰ = 1.07 V

(112)

NO₂ + RH → R^• + HONO

(113)

In aqueous media, NO₂ also hydrolyzes to form nitrous acid and nitric acid via Reaction (114), which is one of the steps in the industrial production of nitric acid from ammonia via the Ostwald process [185].

H₂O
2 NO₂ ⇆ N₂O₄ → HONO + HNO₃

(114)

Although Reaction (114) is negligibly slow at the low concentrations of NO₂ characteristically encountered in the ambient atmosphere, it does proceed upon uptake of NO₂ onto surfaces to produce gaseous HONO in outdoor and indoor environments [186].

Several studies have examined the interactions of NO₂ on the TiO₂ surface under various experimental conditions, such as different NO₂ partial pressures and various temperatures in the range 323–573 K [102,106,108,172]. All these studies reported production of NO in the gas phase, and formation of nitrates on the TiO₂ surface, albeit under photocatalytic conditions.

However, other aspects that seem to have been overlooked by many are the potential specific interactions between the two NO_x molecules and the TiO₂ surface under dark conditions, which need to be re-emphasized constantly.

In this regard, in their 2003 FTIR study carried out in the dark in borosilicate glass vessels, Finlayson-Pitts et al. [186] discovered that the loss of gaseous NO₂ was accompanied by formation of HONO, NO and N₂O; further FTIR studies also revealed the formation of HNO₃, N₂O₄ and NO₂⁺ species, which led them to hypothesize that the symmetric form of the NO₂ dimer, N₂O₄, is taken up on the surface and isomerizes to the asymmetric form, ONONO₂, with the latter undergoing autoionization to NO⁺NO₃⁻. Apparently, it is the latter intermediate species that react with water to generate HONO and surface-adsorbed HNO₃. Subsequently, NO is generated by secondary reactions of HONO on the highly acidic surface. The authors further noted that a key aspect of this chemistry is that in the atmospheric boundary layer where human exposure occurs and many measurements of HONO and related atmospheric constituents (e.g., ozone) are made, a major component for this heterogeneous chemistry is the surface of buildings, roads, soils, vegetation and other materials [186].

A more recent investigation by Sivachandrian and coworkers [187] on the adsorption of NO and NO₂ molecules on the metal oxide TiO₂ at ambient temperature, specifically carried out under dark experimental conditions to avoid any photocatalytic interference, showed no significant adsorption of NO on TiO₂. By contrast, not only did NO₂ significantly influence the adsorption of VOCs and mineralization on the TiO₂ surface, but once the threshold surface coverage of NO₂ was reached at room temperature, the NO₂ adsorbed reactively on the TiO₂ surface by evolving NO in the gas phase. Quantitative measurements performed downstream of the reactor led the authors [187] to propose a new mechanism expressed by the Reactions (115)–(118) for the adsorption of NO₂ on TiO₂ at room temperature under dry air conditions:

2 NO_2(ads) ⇆ NO₂⁻ + NO⁺_(ads)

(115)

NO + O²⁻(TiO_{2 Surface latice}) ⇆ NO₂⁻_(ads)

(116)

NO₂⁻_(ads) + NO_2(ads) ⇆ NO₃⁻_(ads) + NO_(g)

(117)

According to this sequence, the global reaction of NO₂ adsorption is then (Reaction (118)):

3 NO_2(ads) ⇆ 2 NO₃⁻_(ads) + NO_(g)

(118)

Accordingly, the proposed NO₂ adsorption mechanism on TiO₂ at room temperature in the dark [187], together with the experimental observations, could be summarized thus: (i) three NO₂ molecules adsorb on TiO₂, produce two NO₃⁻ ions on the TiO₂ surface, and evolve one NO molecule in the gas phase; (ii) the ratio between consumed NO₂, TPD desorbed NO₂ subsequent to adsorption, and NO produced during NO₂ adsorption = 3:2:1; (iii) the NO₂ adsorption time (i.e., the TiO₂ surface coverage) significantly modified the nature of the adsorbed species at ambient temperature; (iv) the NO formation time was controlled principally by the surface coverage of NO₂⁻ and NO₃⁻ ions, rather than by the NO₂ inlet concentration; and (v) at higher NO₂ gas phase concentrations (greater than 35 ppm) at room temperature, the total amount of consumed NO₂ decreased as a result of self-poisoning of the sites by adsorbed NO₃⁻ species.

Similarly, Haubrich and coworkers [188] examined the interaction of NO₂ on rutile TiO₂(110) and discovered that the presence of NO₂ and water led to the formation of multilayers under dark conditions of nitric acid, HNO₃, contrary to exposure of the surface to pure water after saturation of the surface with 200 mTorr of NO₂; no further growth of the AP-XPS (ambient pressure X-ray photoelectron spectroscopy) nitrate signals occurred under the latter conditions. Apparently, formation of HNO₃ requires weakly adsorbed NO₂ molecules, an important finding with important implications in environmental processes since their study [188] confirmed that metal oxides facilitate the formation of nitric acid under ambient humidity (in the dark) conditions typically encountered in atmospheric environments.

It is evident that there is much more that needs to be investigated to fully understand whatever events occur in and on the TiO₂ (and others) semiconductor photocatalyst. Laboratory experiments using models and solar simulators are just the beginning, after which what is needed is to bring the laboratory outdoors using actual environmental quantities of NO_x as the reagents and humid air as prevails in the environment being investigated for application of the photocatalytic substrates. A careful study of the levels of NO_x or its products at various distances from the photocatalytic surface, both vertically and horizontally, without precluding determination of all other environmental factors is also needed. The deNO_x index has shown that metal-doped TiO₂ systems are also worth investigating further.