The basic principle of photocatalysis was first discovered five decades ago by Fujishima and Honda [28], and the possibilities for using titanium dioxide (TiO₂)-based photocatalysis for removal of various pollutants have since been the subject of various studies [29] [30] [31]. The photocatalytic properties of TiO₂ are based on the semi-conducting properties of the crystalline form of the mineral. TiO₂ nanoparticle exists in three polymorphs, anatase, rutile and brookite. However, the anatase form is the most widely used for photocatalytic purposes [7]. The complex process of NO_x absorption and degradation at the TiO₂ surface is described in detail elsewhere [32] [33]. Illuminated by UV light, the anatase in contact with air and water produces hydroxyl radicals OH∙, which oxidize NO₂ to nitrate ( NO 3 − ):

NO + 2 OH ⋅ → NO 2 + H 2 O (1)

NO 2 + H 2 O → NO 3 − + H + (2)

The resulting NO 3 − is extremely soluble in water; hence it is easily washed away. Since the main catalytic property of the TiO₂ surface is the effective formation of OH radicals, the levels of ozone (O₃), volatile organic compounds (VOCs), and sulphur oxide (SO_x) in the air may also be affected by the contact to the catalytic surface [34] [35] [36]. However, Mothes et al. [37] did not find photocatalytic degradation of the studied toluene and isoprene.

We establish the background data for the LCI for the product NOxOFF supplied by Photocat (Denmark) and consider its application as part of a pre-fabricated concrete element used as pavement and as a sprayed coating on asphalt road.

The production of the NOxOFF products is based on commercial raw materials that are ground and mixed; a SiO₂ product serves as a binder material and ammonia acts as a buffer to stabilize pH. Production of 1 kg of product is summarized in Table 1.

The product is transported to the application in black plastic containers (IBC) holding 1000 kg. The container consists of high density polyethylene (HDPE) and weighs 59 kg. The containers are used 5 - 10 times and afterwards sent to a municipal waste incinerator with energy recovery.

When used in concrete elements, the product is introduced during the manufacturing of the concrete product before the product is cured and the NOxOFF catalyst is integrated into the top 2 mm of the final product used for the final surfaces.

When used on asphalt roads and parking lots, the product is sprayed applied when the initial surface bitumen is worn off and the NOxOFF product can adhere and bind to the mineral granulates in the surface of the asphalt. The product is spray-applied on existing asphalt pavements using common equipment. We neglect any environmental burdens from the application process.

Modelling of operation of photocatalytic surfaces is based on full-scale experiences with the photocatalytic surface has been tested during the last 15 years [11] [38]. Pedersen et al. [11] proposed a scheme for evaluation the different real-world studies including lifetime and durability. The lifetime not only depends on the quality of the coating but also on the quality of the material to which the coating adheres. We estimate that the lifetime of the photocatalytic concrete pavement elements is 15 years and in case of the asphalt road 10 years. We do not expect any maintenance during the lifetime and we do not expect that the photocatalytic surface coating affects other kinds of maintenance that may routinely take place on the surfaces. Faraldos et al. [39] and Bendix et al. [40] showed that the coating has a cleaning effect on the surface but we have not been able to quantify this additional effect into a format suitable for the LCA.

For the NOxOFF product, three major real-world demonstrations were performed during 2012-2019 [41]. From the study in the city of Roskilde in Denmark, it was calculated that the NOxOFF product removed 13.8 g of NO_x/m² per year. This was based on laboratory ISO test results, which were correlated to real-life results with good approximation. The tests showed a NO_x removal efficiency of approximately 4.83 mg/m² per hour. The UV influx in Denmark is around 50 kWh/m² per year. NO 3 − is the oxidized product. This means that 10 m² of coated surface during a 10 year lifetime removes 1.38 kg NO_x and produces around 2.65 kg NO 3 − . We assume that the NO 3 − is dissolved into water and that the associated acidity production is only of marginal significance in the water that washes the surface during the lifetime of the product. We take into account the produced NO 3 − and, although the destination may vary, we assume that marine surface water will be ultimately affected.

In the LCA modelling we also considered effects on other air pollutants. Baral et al. [16] as well as Hernández-Alonso et al. [42] quantified the removal of SO_x by photocatalytic surfaces. By scaling the removals of SO_x and NO_x observed over a 15 year life time [16], we include a removal of 0.033 kg SO_x per kg NO_x removed. The SO_x is converted to sulphate that is washed off. The data on O₃ removal is scarce [36] and not yet quantified relatively to the NO_x removal, hence we did not include it.

The photocatalytic surface is not reused or recycled after its lifetime. We assume that the main product amended with surface coating after the end-of-life of the main product will enter into traditional material recycling and disposal and that the presence of the coating will not affect the environmental aspects of the material recycling and disposal. The photocatalytic effect is assumed lost when the material is being recycled. We assume that the final disposal of the photocatalytic coating occurs together with the concrete and asphalt pavement to which it is bound, and that the environmental profile of the recycling is not affected by the presence of the photocatalytic material.

Based on the above data compilation, the LCI for 1 kg NO_x removed by photocatalytic surface coating has been developed as presented in Table2. Table2 summarizes resources in terms of materials and energy required for each life-cycle phase of the technology. The inventory differs with respect to photocatalytic surface coating applied to concrete or to asphalt pavement. The corresponding Ecoinvent processes utilized for the modelling of photocatalytic surface coating are reported in the SM, TableS1 and TableS2.

The knowledge about the ability of certain surfaces to enhance chemical reactions is very old and has been used in catalytic converters applied as end-of-pipe technologies for vehicles since 1975 in US and since 1985 in Europe [43] [44]. Today new cars are supplied with catalytic converters as emission control technologies. This applies to passenger cars, vans, and trucks as well as to gasoline and diesel driven vehicles. Several types of catalytic converters exist [5], but we have been able to build stringent LCI data only for the TWC (Three-Way Catalyst) which is often used on gasoline-driven passenger cars. Diesel cars and trucks use other types of after treatment systems, for example the Urea-Selective

Catalytic Reduction (Urea-SCR) which uses urea additive to convert NO_x to N₂ and water [45]. The TWC-converter creates reductive and oxidative reactions at elevated temperature that decrease/remove the content of air pollutants in the engine exhaust. Several reactions take place [46] and many air pollutants are affected. The dominating overall catalytic conversions are:

NO x → N 2 + H 2 O (3)

CO → CO 2 (4)

THC → CO 2 + H 2 O (5)

where THC is total hydrocarbons.

The TWC converter has first a reductive section where NO_x is removed and a second oxidative section where CO and THC are removed. During the NO_x reduction, oxygen is generated and hence additional oxidation reactions may take place in the reductive section. Some oxygen will feed into the oxidative section of the converter and contribute to the oxidation of CO and THC. The catalytic converter increases the fuel consumption because of the pressure loss in the converter.

Laboratory test as well as in-situ measurements on vehicles in operation have documented the effect of catalytic converters on vehicle emissions [47], but the studies have also demonstrated that the effect depends on many technical factors as well as temperature and mode of operation [46].

We establish the background data for the LCI for the TWC converter based on information in the literature. Due to the complexity of the reactions and the integrated function of the reductive and the oxidative sections of the converter, we establish two sets of data. A first data set for the whole converter considering the combined effect of the converter, and a second data set for the first part of the converter assuming that the reductive section only removes NO_x. We aim at establishing robust LCIs of general value and, although we rely on specific data from Belcastro [18], we assume that the data can be generally representative of the performance of a generic catalytic converter, irrespective of vehicle, product or driving conditions.

A vehicle catalytic converter is contained in a steel canister and operates typically around 350˚C with a conversion efficiency above 90% [48]. The exhaust gas is routed through two chambers with honeycomb structures with coated surfaces providing large catalytic surfaces sustaining the reactions. The catalytic elements are typically rhodium in the reductive section and platinum and palladium in the oxidative section. However, variations exist in both the construction, the catalyst, and their catalytic metal composition. Typical data are compiled in Table 3.

The cordierite is specifically produced as the honeycomb structure and no database offers an LCI for this product. We use data based on the compilation provided by Belcastro [18] as presented in Table 4 for a Cordierite honeycomb structure

**Data are half the data from the TWC converter except the rare elements, which are distributed according to their function; **Coarse estimates based on data used by Belcastro [18].

at 1400 gram. The weight of input materials exceeds the final weight of the cordierite by approximately 10% because of the water content of the input materials (deducted from Belcastro [18] ).

The production of the canister from the steel requires energy, which we estimate as the energy needed with the energy required for the steel deep drawing process, as assumed by Belcastro [18].

Catalytic converters are typically manufactured based on components produced by specialized industries on commercial raw materials. The converters are in large quantities transported to the car assembly facilities. We do not have specific information on the packaging and transport of the individual components but we assume that the cardboard box for the converter weighs around 1 kg based on information in Belcastro [18].

As far as the operation phase is concerned, we assume that a converter has a lifetime corresponding to 160,000 km driven [49] [50]. The converter is estimated to increase the fuel consumption due to lower air-to-fuel ratio and back pressure by 2% - 10% depending on the driving conditions [51]. Assuming an emission of 150 - 220 g CO₂ per km for a typical gasoline-driven passenger car [47], we estimate that the additional fuel consumption for the TWC converter amounts to approximately 2000 kg CO₂. We assume that the converters require no maintenance during their lifetime.

Estimating the amount of NO_x and other air pollutants removed during the lifetime of the converter requires data on average composition of the exhaust gas before and after the converter.

Several studies have addressed the removal of air pollutants by catalytic converters [46], but, since the engine technology as well as the catalytic converters constantly develop, only recent data should be applied. Fanick et al. [52] measured the exhaust gas composition before and after the converter on two Toyota models. The results are summarized in SI units in Table 5. As baseline for the LCA modelling over the TWC lifetime, we use a removal of 250 kg NO_x, 1150 kg CO, 180 kg THC. In addition, the CO₂ emissions increase with 2350 kg due to the conversion of THC and CO into CO₂.

For the end-of-life phase of vehicle catalytic converters, we assumed that the majority of the converters are dismantled and properly managed. The energy

and material requirements for the treatment of spent catalytic converters were taken from the Ecoinvent database. Platinum, rhodium and palladium may be recycled [53], and the remaining parts are landfilled. We considered two scenarios, one where the metals are not recycled, and one where the metals can be fully recycled, and later discuss how this affects in the LCA results for the catalytic converter.

Based on the above data compilation, the LCI for 1 kg NO_x removed by the catalytic car converter has been developed as presented in Table6. Table6 summarizes resources in terms of materials and emission involved in each life-cycle phase of the technology. The inventory column TWC represents a whole converter including both the first part where the NO_x reduction take place and the second part where the reduction of THC and CO primarily takes place. The inventory column 1/2 TWC represents only the first part of the converter and is ascribed the NO_x removal, the use of rhodium plus a proportional part of the materials for the honeycomb and the canister as well as the additional fuel use. This approach is used to bracket the environmental profile of the NO_x removal, since the processes cannot be separated as stringently as done here. The corresponding Ecoinvent processes utilized for the modelling of catalytic car converter are reported in the SM, TableS3 and TableS4.

The direct effect of the NO_x removal dominates the environmental profile of the photocatalytic coatings, showing significant savings in POF, A, and TE ranging approximately from −12 to −24 mPE per kg NO_x. Some savings are also seen in PM (−2 mPE per kg NO_x). The net value for ME is close to zero; this reflects that the NO_x, with and without the photocatalytic surface, always ends up as NO 3 − .

The burdens from the production of the coatings are minor. Only a small burden of approximately 1 mPE per kg NO_x is observed in FET due to the provision of TiO₂.

The associated benefits of removal of SO_x are marginal and do not provide a large contribution to the results (not visible contribution in Figure 3). No significant impacts are seen from the end-of-life phase.

The overall results obtained for the photocatalytic coatings showing that the benefits of removing the NO_x far outbalances any environmental impacts from the production of the coating and is in accordance with what has been the main outcome of a few previous studies [15] [16]. The main environmental impact connected to the production phase is due to the provision of TiO₂. We include accurately the main materials and process in the production of the coating, and we follow standard procedures in quantifying the benefits of removing the NO_x. Thus, we find the overall result robust, although we cannot judge their general representability due to the few data sets available. However, we may not fully cover the additional benefits of using the photocatalytic coating on surfaces. This is primarily due to lack of clear evidence and data. SO_x removal seems of no importance for the results, but we neglect any effects related to removal of O₃, oxidation of volatile hydrocarbons, effects related to the cleanness of the surface, and the potential increase in uptake of CO₂ in the concrete due to the presence of TiO₂. These aspects have been addressed in the literature [36] [40] [54] [55], but are all still very new and characterization factors and methods to account for these aspects in an LCA are still not available. However, although not possible to quantify at the moment, we believe that these additional functions (removal of O₃, oxidation of VOCs, effects related to the cleanness of the surface, and the potential increase in uptake of CO₂) all potentially would increase the benefits of reduction of NO_x via catalytic surface coating.

The direct effect of the NO_x removal dominates the environmental profile also of the catalytic converter. The effects in PM, POF, A, and TE are identical to the benefits also seen for removal of 1 kg NO_x by the photocatalytic surface coating, but the converter also shows a significant saving in ME (−13 mPE per kg NO_x). This is due to the fact that the catalytic converter produces N₂ and not NO 3 − , and N₂ has no environmental impacts. In addition, the TWC shows additional benefits in HTNC (10 mPE per kg NO_x) and in POF (5 mPE per kg NO_x). These additional benefits are not observed for the 1/2 TWC; they are caused by THC and CO reduction that primarily takes place in the second half of the converter, while NO_x removal takes place in the first part of the converter.

The converter also reveals a minor burden in CC (2 mPE per kg NO_x), due to the additional CO₂ emissions from the conversion of THC and CO to CO₂ and from the additional fuel use. Overall, the CC impacts connected to the provision of materials and energy for the NOx removal function of the converter is only 1/4 of the CC impact for the whole converter.

The burdens from the production of the converter are significant in the FET (TWC: 19 mPE per kg NO_x; 1/2 TWC: 7 mPE per kg NO_x). The provisions of the rare earth elements for the catalytic honeycomb coating are the cause. The NO_x removal takes place in the first part of the converter where only rhodium is used (7 mPE per kg NO_x), while platinum and palladium are used only in the second part of the converter (12 mPE per kg NO_x). The other burdens for the production are very small (A and FE: <1 mPE per kg NO_x). The end-of-life shows a small burden in RDM: around 1 mPE per kg NO_x), due to the management of the spent converter.

Recycling of rare earth elements from the spent converter is not included in the main results because we have no reliable data on how frequently the converters are recycled nor on the efficiency of the recycling process regarding the actual recovery of the individual elements. Work is in progress in this area [56], and when data are established this impact can be reduced proportionally to actual recovery of each metal. A sensitivity analysis considering potential full recovery of metals shows that impacts connected to the provision of rhodium, palladium and platinum is counter-balanced by their full recovery as secondary raw materials, which reduces the impacts for the FET to <1 mPE.

The overall results obtained for the vehicle catalytic converter showing that the benefits of removing the NO_x well outbalance the environmental impacts from the production of the converter, is in accordance with what has been the main outcome of a few previous studies. Nevertheless, different approaches and assessments methods make direct comparison difficult [17] [18] [57].

The fact that the benefits of the NO_x removal have a relatively higher contribution to the overall results and the fact that the NO_x removal even may claim some of the benefits from the removal of THC and CO (since the NO_x removal provides oxygen for the second part of the converter), we believe that the overall results are robust, although several assumptions were made regarding the production of the converter and several proxies were use where datasets were missing.

The impacts from the production of the converter are related to the use of rare earth elements. The fact that we did not consider recovery of the rare earth elements, may mean that we overestimate the environmental burden from producing the converter. Recovery will reduce the impact proportional to the recovery efficiency because the burden is linked to the mining and refinement of virgin resources. Some of the rare earth elements are produced as by-products in other mining operations and the allocation of environmental burdens for virgin materials as well as the environmental savings form recovery depend on the development in the markets both in terms of supplies and prices, but it is likely that they are related, and thus recovery always will reduce the burden of producing the converter.