Predictive Modeling of Air Purification Efficiency in Nano-TiO₂-Modified Photocatalytic Cementitious Composites Using High-Resolution EDS Mapping and Mercury Intrusion Porosimetry

Abstract

This study investigates the relationship between surface properties and microstructural characteristics of photocatalytic composites and their impact on air purification efficiency. High-resolution energy-dispersive X-ray spectroscopy (EDS) mapping and mercury intrusion porosimetry (MIP) were employed to analyze photocatalyst distribution and pore structure quantitatively. The findings demonstrated a strong correlation between TiO₂ coverage on the photoactive surface and NO removal rates and between pore structure characteristics and NO₂ generation rates. Two predictive models were developed to link NOx removal rates with photocatalytic cementitious mortars’ surface and structural properties. A stepwise regression approach produced a second-degree polynomial model with an adjusted R² of 0.98 and a Mean Absolute Percentage Error (MAPE) of 8.34%, indicating high predictive accuracy. The results underscore the critical role of uniform photocatalyst distribution and optimized pore structure in enhancing NOx removal efficiency while promoting the generation of desirable products (NO₃⁻) and minimizing the formation of undesirable byproducts (NO₂).

1. Introduction

Air pollution is one of the most critical global challenges, profoundly affecting public health, environmental stability, and climate. Key pollutants such as nitrogen oxides (NOx), sulfur oxides (SOx), volatile organic compounds (VOCs), carbon monoxide (CO), and particulate matter (PM) predominantly arise from human activities, including industrial processes, transportation, and energy production. These pollutants have severe health impacts, including respiratory and cardiovascular diseases. According to Beng [1], exposure to air pollution was linked with 9 million premature deaths worldwide in 2019.

Photocatalytic materials offer a promising urban passive air purification solution, using light-induced chemical reactions to degrade pollutants. Under specific light wavelengths, reactive free radicals (primarily hydroxyl radicals (^•OH) and superoxide anions (O₂^−•)) are generated on the surface of photocatalytic grains, oxidizing pollutants from the external environment, converting harmful gases like NOx into nitrates (NO₃⁻), or breaking down hydrocarbons into byproducts such as CO₂ and H₂O. The intensity and effectiveness of the aforementioned reactions depend on numerous factors extensively studied over the years [2,3,4]. Various environmental variables are crucial in determining their efficacy when evaluating the air purification potential of photocatalytic materials. Among these, the concentration of airborne pollutants, the intensity and wavelength of electromagnetic radiation [5,6], ambient relative humidity [7], temperature conditions within the reaction environment, and the dynamics of airflow [8] have been identified as statistically significant factors that influence photocatalytic performance. Furthermore, when conventional construction materials such as concrete and other cementitious composites are modified with photocatalytic additives, these composites’ physical and chemical properties exert additional influence on their interaction with environmental factors (Figure 1).

The accessibility of air pollutants to photocatalysts is a critical factor influencing the effectiveness of the photocatalytic process [9,10]. Since photocatalyst grains are typically embedded within an organic or inorganic matrix—either as a photocatalytic coating on a non-photocatalytic material or as a material directly modified with photocatalysts—the properties of the photoactive surface play a significant role in regulating the interaction between photocatalysts and pollutants [11,12]. So far, several surface characteristics that significantly influence the performance of photocatalytic cementitious materials have been indicated—surface texture, pore structure, and distribution of photocatalysts [12,13].

The contact area between pollutants and the photoactive surface can be influenced by modifying the macro- and microtexture of the surface. Roughening the surface has been shown to enhance photocatalytic performance. For instance, Boonen and Beeldens [14] reported a threefold improvement in NO removal rates for sawn surfaces compared to smooth formwork surfaces. Conversely, Guo et al. [15] found no significant change in the photocatalytic performance of cement mortars after 500 abrasion cycles, suggesting that the degree of roughness is a crucial factor. Furthermore, in their earlier study [3], the authors demonstrated that increasing the roughness of a photocatalytic surface can help mitigate reductions in photocatalytic performance when coarse aggregate is exposed.

Higher porosity improves pollutant diffusion into the composite, boosting the air purification performance. Zhang et al. [16] reported a 115% increase in NO removal rates as the porosity of cement-based composites rose from 26.5% to 61.4%. Similarly, Hamdany et al. [17] highlighted that increasing total porosity enhances photocatalytic performance by providing a greater accessible active surface area. However, recent research conducted by the authors [18] emphasized that overall porosity and pore structure play a significant role in determining air purification efficiency. Their findings demonstrated that an increase in the content of capillary pores (0.1–1.0 µm) substantially improved the selectivity of NOx removal.

A uniform dispersion of photocatalysts, such as titanium dioxide (TiO₂), enhances pollutant degradation by increasing the active surface area and ensuring consistent light absorption. Research by Wei et al. [19] emphasized that optimizing the distribution of photocatalytic nanoparticles on building materials significantly improves the removal rates of airborne pollutants, highlighting the importance of controlled deposition techniques. Similarly, Lorencik et al. [20] observed that coatings with a more uniform distribution of photocatalysts demonstrated significantly better performance in air purification tests.

2. Research Significance and Framework

Energy-dispersive X-ray spectroscopy (EDS) mapping is widely used to evaluate the distribution of photocatalysts on photoactive surfaces. This technique provides detailed spatial information about the surface’s elemental composition, making it particularly valuable for assessing the uniformity and quality of photocatalyst dispersion. EDS mapping data often correlate with photocatalytic performance tests to explore the relationship between photocatalyst distribution and pollutant degradation efficiency [21,22].

However, to date, no comprehensive analysis has been conducted to assess the potential of this technique in predicting air purification performance across a broader range of photocatalytic cementitious materials. To address this gap, the authors aimed to evaluate the usefulness of EDS mapping in modeling the air purification performance of cement-based photocatalytic mortars.

In this study, the authors employed high-resolution EDS mapping combined with image analysis algorithms to quantify the distribution of photocatalysts on the surface of cement composites. Additionally, mercury intrusion porosimetry (MIP) was used to investigate the cross-effects of pore structure on air purification performance, providing a holistic approach to understanding the interplay between material properties and functional performance.

Twelve specimens with significantly different air purification performances yet similar material compositions were tested in this study. The objective was to investigate whether the pore structure and/or the distribution of photocatalysts on the surface influence their photocatalytic performance. Six different mortar compositions were evaluated, incorporating two types of photocatalysts and varying key parameters such as the water-to-cement ratio, cement-to-sand ratio, photocatalyst content, and the mass ratio of one photocatalyst to the other (

Table 1. The composition of mortars evaluated in this study (acronyms: w/c—water-to-cement ratio; c/s—cement-to-sand ratio; all ratios were expressed as mass ratios).

Mortar ID	w/c [-]	c/s [-]	Total Photocatalyst Content [kg/m3]	K7000/P25 Ratio [-]
PCM-1	0.40	0.80	12.5	0.625
PCM-2		0.68	12.5	1.000
PCM-3		0.68	15.0	0.250
PCM-4		0.68	15.0	0.625
PCM-5		0.80	12.5	0.250
PCM-6		0.80	15.0	0.250

). The photocatalysts were added in powder form during the mixing process.

All tested samples were subjected to the same preparation and curing regime. The samples of dimensions 140 × 160 × 40 mm were covered with plastic foil for 24 h before demolding. After demolding, they were cured for 28 days in a curing chamber (T = 20 ± 2 °C; RH > 95%) before further testing. Following curing, the samples were subjected to an air purification performance test. Subsequently, a smaller sample measuring 20 × 20 × 10 mm was cut from the center of each 140 × 160 × 40 mm specimen for additional analysis, including EDS mapping and MIP analysis (Figure 2).

3. Materials and Methods

3.1. Materials

Cement CEM I 42.5 R, conforming to the PN-EN 197-1 [23] standard (Ożarów, Poland), was utilized to prepare the investigated cement mortars. Its mechanical properties were verified following the PN-EN 196-1 [24] standard, including assessments of initial and final setting times in accordance with PN-EN 196-3 [25]. Additionally, its specific gravity, chemical composition, and specific surface area were determined using the Blaine method—

Table 2. Characteristics and chemical composition of the binder used in this study.

CEM I 42.5 R	Flexural strength, MPa		Compressive strength, MPa			Initial setting time, min		Final setting time, min		Specific gravity, g/cm3		Specific surface area, cm2/g
	4.46		44.4			190		270		3.09		3920
	Chemical composition, % of mass
	CaO	SiO2		Al2O3	MgO		K2O		Na2O	Fe2O3	P2O5		LOI
	61.4	21.7		5.2	2.6		0.7		0.2	2.4	0.2		2.3

.

This study included a combination of two photocatalytic materials, representing first- and second-generation titanium dioxide (TiO₂)-FG (P25, Shanghai, China) and (TiO₂)-SG (K7000, Leverkusen, Germany), respectively—Figure 3. The properties of these materials were detailed in [26]. Key characteristics, including the content of individual crystalline phases, crystallite size, and specific surface area, are summarized in

Table 3. Physical characteristics of selected photocatalytic nanomaterials, including their specific surface area (BET), crystalline phase content (XRD), and diameter of individual crystallites (XRD).

Photocatalytic Material	Specific Surface, m2/g	Crystallite Diameter, nm		Crystalline Phase, %
		Anatase	Rutile	Anatase	Rutile
(TiO2)-FG	53.8 ± 0.2	33	54	87	13
(TiO2)-SG	246.8 ± 2.9	10	-	100	-

. The mass ratio of the two photocatalysts in the investigated cementitious composites varied (TiO₂)-SG to (TiO₂)-FG = 0.25–1.0 [-]), as did the total TiO₂ content (12.5 or 15 kg/m³). Combining these photocatalytic materials was selected to enhance the composite’s photocatalytic activity across a broader light spectrum. First-generation TiO₂ (P25) exhibits peak activity within the UV-A range. At the same time, the second-generation TiO₂ (K7000) extends the activation band into the visible light spectrum due to the introduced carbon complex attached to the surface of K7000 nanoparticles, leading to adsorption in the visible light spectrum [27]. Both photocatalytic materials were characterized by regular grain morphology.

As an aggregate, fire-dried quartz of properties presented in

Table 4. Selected characteristics of fire-dried fine aggregate used in this study.

Characteristics	Fire-Dried Fine Aggregate 0.1/0.5	Fire-Dried Aggregate 0.5/1.2
Dust content, %	0.1
Water absorption, %	0.1
SiO2 content, %	99.6	99.4
Sand equivalent, -	99.06	99.17
Specific gravity, g/cm2	2.65

and of granulation 0.1/0.5 and 0.5/1.2, meeting the requirement of EN 13139 [28], was used. The proportion between the mass content of aggregates of different granulations was kept constant between the investigated mortar series and equaled 1.

An aqueous polycarboxylate superplasticizer (HES 1517, Myślenice, Poland) was included in the composition of cement mortars to modify their rheological properties. It met the requirements of PN-EN 934-2 [29] and was characterized by electrostatic and steric mechanisms of action. To keep the consistency of investigated mortars at the same level, its content varied between individual series in the range of 0.3 to 0.6% (consistency of 26 to 28 cm measured via flow table method).

3.2. Methods

The evaluation of air purification efficiency by photocatalytic cementitious composites in removing gaseous pollutants, specifically nitrogen oxides (NOx), was conducted using a test methodology developed by the authors and aligned with the ISO 22197 standard [30]. A cuboid sample measuring 140 mm × 160 mm × 40 mm was prepared for each mortar series under investigation. Following casting, the samples were demolded and subjected to a 28-day curing period in a controlled curing chamber maintained at a relative humidity exceeding 95% and a temperature of 20 ± 2 °C. Notably, the curing process excluded water immersion to avoid potential alterations in the surface characteristics of the composites, such as efflorescence or the deposition of precipitates of unknown origin. These surface changes could adversely affect the photocatalytic performance of the mortar’s photoactive external layer, compromising the experimental results’ reliability and reproducibility.

Before conducting the air purification tests, the samples underwent a meticulous preparation process to ensure the cleanliness and integrity of the photoactive surfaces. Initially, the external surfaces were thoroughly cleaned with distilled water and gentle scrubbing to remove visible macro-contaminants. Subsequently, the samples were dried at 60 °C for two hours. Following the drying phase, they were transferred to an irradiation chamber and exposed to UV-A light at 10 W/m² for 16 h. This step was designed to thermally and photochemically degrade any residual organic impurities on the photoactive surfaces. After irradiation, the samples were rinsed again with distilled water to remove any loose particles or byproducts of the aforementioned decomposition. They were dried at 60 °C for a final time. The air purification efficiency, specifically targeting NOx removal, was assessed no earlier than two hours after the completion of the preparation process. This timeline was implemented to stabilize the surface characteristics and ensure consistent conditions for the photocatalytic activity tests. This rigorous sample preparation protocol aimed to standardize the experimental conditions and enhance the reliability of the observed results.

The experimental evaluation of the photocatalytic efficiency of cementitious samples was conducted in a setup ensuring controlled and reproducible conditions. The prepared samples were placed inside a sealed glass reaction chamber with a total internal volume of 4 dm³. The photoactive surface of each sample was oriented to face the light source directly, ensuring maximum exposure to the incident radiation (Figure 4).

The gas flow rate through the chamber was regulated at 2 L per minute, with a nitrogen oxide (NOx) concentration at the chamber inlet maintained at 100 ± 5 ppb. This concentration was chosen based on the average nitric oxide levels recorded by the monitoring stations in Warsaw, ensuring that the experimental setup closely replicates real-life pollution scenarios. The results provide a meaningful evaluation of the photocatalytic composite’s performance in practical applications by aligning the test conditions with typical urban NOx exposure.

To replicate standardized environmental conditions, the test was conducted in an ambient air environment, with temperature and relative humidity closely monitored and maintained at 20 ± 2 °C and 40 ± 5% RH, respectively.

The lighting system used in the tests comprised two low-radiation LED sources, carefully chosen to simulate typical lighting conditions prevalent in Poland during the autumn and winter. These included a UV-A light source (365 ± 5 nm, 14.4 W/m, MEISSA, Warsaw, Poland) emitting ultraviolet radiation at a wavelength of 365 nm and an intensity of 1.0 W/m² and a visible light source (25.0 W/m, Sun-Like™ TRI-R™ LED 5000 K, Toshiba Materials, Tokyo, Japan) providing global radiation at 150 W/m². Notably, the visible light source was engineered to exclude radiation below 400 nm, ensuring that any photocatalytic activity observed under this condition was attributable solely to visible light.

The concentrations of nitrogen monoxide (NO) and total nitrogen oxides (NOx) within the chamber were continuously measured throughout the experiment using the Teledyne API T200 chemiluminescence detection analyzer (San Diego, CA, USA).

The experimental procedure was systematically divided into several sequential stages to investigate the effect of different irradiation conditions on photocatalytic performance (Figure 5). Initially, the NOx concentration in the reactor was stabilized at a constant level of 100 ± 5 ppb. Thereafter, the samples were exposed to alternating lighting conditions to simulate various irradiation scenarios: visible light only, UV-A light only, and a combination of both visible and UV-A light. Each phase was carefully monitored to assess the impact of individual and combined light spectra on the degradation of NOx.

Each irradiation phase lasted 30 min, during which the specific light source(s) were continuously active to illuminate the photocatalytic surface. Between active irradiation phases, 30 min intervals were incorporated where no light source was active. These intervals established a baseline condition and allowed the system to equilibrate, thereby isolating the effects of each lighting scenario on the photocatalytic activity of the cementitious samples.

During each irradiation phase, the concentrations of nitrogen monoxide (NO), nitrogen dioxide (NO₂), and the total nitrogen oxides (NOx) in the reaction chamber were monitored. The stabilized NO concentration in the example presented in Figure 5, before the first cycle of light activation, was approximately 105 ppb. This stabilization was determined by a variation of no more than 0.2 ppb across 20 consecutive 10 s measurements. The first cycle began once this stable concentration was reached. In subsequent cycles, after the light was turned off, the NO concentration increased and stabilized again at approximately 105 ppb, reflecting the equilibrium state of the chamber in the absence of photocatalytic activity. Simultaneously, a decrease in NO₂ concentration was observed due to the cessation of photocatalytic reactions, which are light-dependent and necessary for producing NO₂.

Key parameters were quantified for each lighting condition, including the amount of NO introduced into the reactor, the quantity of NO removed via photocatalytic reactions, and the NO₂ generated as a byproduct of the process. The removal and generation rates were subsequently calculated and expressed as the absolute reduction or production of NOx in micrograms per hour per square meter of the photocatalytic cementitious surface (µg/h × m²). This approach allowed for a standardized comparison of photocatalytic efficiency under distinct irradiation scenarios.

Furthermore, the selectivity of NOx removal—a critical parameter indicating the efficiency of converting NO into NO₂ or other end products—was calculated using a quantitative approach. The selectivity metric (S) was derived from the ratio of NO mass removed to NO₂ mass produced during the photocatalytic reactions (Equation (1)). This was determined based on the integrated area measurements highlighted in Figure 5, corresponding to the variations in NO and NO₂ concentrations over time.

$$\mathrm{S}=1-{\displaystyle \frac{{generatedNO}_2}{removedNO}}$$

(1)

The distribution of photocatalysts on the surfaces of the mortar samples was investigated using scanning electron microscopy (SEM) combined with energy-dispersive X-ray spectroscopy (EDS). This analytical technique enabled detailed visualization and compositional analysis at the microstructural level. For this study, approximately 20 mm × 20 mm × 10 mm samples were sectioned from the central region of the original mortar specimens, which initially measured 40 mm × 40 mm × 160 mm. The central sectioning ensured a representative assessment of the photocatalyst distribution, minimizing edge effects or other anomalies.

The EDS analysis specifically targeted the spatial distribution of titanium, a primary constituent of the photocatalyst. A systematic graphic processing protocol was applied to the raw data to enhance the accuracy of the titanium distribution maps. Weak signal points, which could originate from titanium within the binder matrix or aggregate inclusions, were excluded from the analysis. This step was essential to ensure that only the titanium associated with the active photocatalytic layer was considered. Processed images were subsequently binarized, enabling clear differentiation between regions with and without detectable titanium signals. Pixel counts corresponding to titanium-rich areas were quantified through histogram analysis, providing a measure of photocatalyst distribution.

A standardized methodology was employed to process and analyze the EDS maps. Initially, each map, inherently composed of RGB pixel data, underwent monochromatic conversion. This step simplified subsequent image processing while retaining the essential spatial information. A static binarization threshold was applied uniformly across all maps. This threshold was selected based on preliminary calibration to optimize the contrast between photocatalyst (high titanium signal) and non-photocatalyst (background noise) regions. The binarized maps delineated photocatalyst distribution (Figure 6).

To comprehensively evaluate the photocatalyst content and distribution, 42 EDS maps were generated for each sample type, resulting in 504 EDS maps across all sample types. Each map covered a micro-area measuring 0.31 mm × 0.24 mm, chosen to balance resolution with representative surface coverage. By aggregating data from 7 × 6 maps per sample, a total analyzed area of 2.17 mm × 1.44 mm was obtained for each sample. These 42 merged maps were combined into a single high-resolution EDS mapping binarized using a static threshold of 67.

For the photocatalyst uniformity analysis, each binarized high-resolution mapping was divided into a 17 × 12 square grid, with each square corresponding to an area of 52 × 50 pixels. The TiO₂ coverage was calculated for each square, and the resulting data were analyzed using statistical methods. Heatmaps were generated to visualize the spatial distribution, and the Chi-Square Goodness-of-Fit Test and Spatial Autocorrelation Tests (Moran’s I and Geary’s C) were applied with a significance level of 0.05. These analyses provided a quantitative assessment of the uniformity and spatial distribution of the photocatalyst across the sample surfaces.

The Chi-Square Goodness-of-Fit Test quantified the deviation between observed values (white pixel area coverage in each grid cell) and expected values (ideal uniform distribution). This test determined whether the photocatalyst distribution was statistically uniform. The null hypothesis (H0) assumed that the photocatalyst (white pixels) was uniformly distributed across all grid cells, with no significant variation in coverage. If H0 was rejected (p < 0.05), it indicated that the distribution of white pixels was non-uniform, highlighting patterns such as clustering or uneven distribution of the photocatalyst.

To further analyze the spatial characteristics of the distribution, Spatial Autocorrelation Tests were performed, specifically Moran’s I and Geary’s C. These metrics assessed whether the photocatalyst distribution was random, clustered, or dispersed across the grid. Spatial autocorrelation analysis is valuable for understanding spatial relationships within a grid and detecting patterns such as clustering, dispersion, or randomness in the data.

Moran’s I is a global spatial autocorrelation metric, measuring the overall clustering of similar values across the grid. Moran’s I values range from −1 to 1, as follows:

• I > 0: positive spatial autocorrelation, indicating clustering of similar values.
• I = 0: no spatial autocorrelation, suggesting a random distribution.
• I < 0: negative spatial autocorrelation, indicating a dispersed or checkerboard-like pattern.

Geary’s C, on the other hand, focuses on local spatial autocorrelation and is more sensitive to local differences between neighboring grid cells. Geary’s C values range as follows:

• C ≈ 1: random spatial distribution.
• C < 1: clustering of similar values, indicating local autocorrelation.
• C > 1: dispersion or dissimilarity between neighboring cells.

These spatial autocorrelation metrics provided a comprehensive evaluation of the photocatalyst distribution. Moran’s I revealed the global spatial patterns of clustering or randomness, while Geary’s C offered insights into local variations and the relationships between neighboring grid cells. Together, these methods ensured a detailed understanding of the spatial distribution characteristics of the photocatalyst.

Microstructural imaging was performed under low vacuum conditions (80 Pa) using a field emission scanning electron microscope (Nova NanoSEM 200, FEI, Hillsboro, OR, USA) to ensure high-resolution surface characterization while minimizing sample charging. An acceleration voltage of 15 kV was applied during imaging, providing optimal penetration and resolution for observing fine microstructural details. Energy-dispersive X-ray spectroscopy (EDS) was employed to enhance the analysis further, utilizing an Octane Elect detector (EDAX) to acquire detailed elemental distribution maps. This combination of SEM and EDS allowed for precise correlation between the morphological features and compositional data of the materials.

The porosity structure of hardened mortars, particularly their near-surface pore characteristics, was analyzed using mercury intrusion porosimetry (MIP) with an Autopore IV 9510 porosimeter (Micromeritics, Atlanta, GA, USA). This technique provided detailed quantitative insights into the pore size distribution, total porosity, and pore connectivity of the cementitious samples. The analysis focused on characterizing pores within a size range spanning approximately 0.003 to 360 μm, covering micro- and mesopores critical to the material’s transport and durability properties. Specimens measuring 10 mm × 10 mm × 10 mm were mechanically sectioned from the photoactive surface of larger hardened mortar prisms of 40 mm × 40 mm × 160 mm. To ensure uniform curing and hydration, the original samples were submerged in water for 28 days under controlled conditions. The specimens were mechanically cut to the required dimensions following this curing phase. The smaller samples were then oven-dried at 105 °C for 48 h to eliminate residual moisture, which could otherwise interfere with the mercury intrusion process by impeding accurate pressure measurements or altering pore wall characteristics.

The mercury intrusion technique is based on the application of external pressure to force non-wetting mercury into the pores of the material. The relationship between the applied pressure and the corresponding pore size is described by the Washburn equation (Equation (2)), where r—pore radius [m], γ—surface tension of the liquid [N/m], θ—contact angle between the liquid and the solid surface [rad], and P—applied pressure or capillary pressure [Pa]:

$$r={\displaystyle \frac{-2\gamma \cos \theta }{P}}$$

(2)

The porosimetry analysis yielded data regarding the cumulative pore volume, the distribution of pore sizes, and the threshold pore diameter, representing the transition between interconnected and isolated pore networks. These parameters are crucial for understanding the transport phenomena within cementitious materials, such as capillary absorption, gas diffusion, and ion migration, significantly influencing durability and performance. By focusing on near-surface porosity, this investigation provided insights into the interaction between environmental exposure and the material’s photocatalytic activity, as surface pore characteristics directly affect light penetration, pollutant accessibility, and overall reactivity.

4. Results

4.1. Air Purification Performance

The photocatalytic performance of investigated mortar samples displayed considerable variability across key parameters, including the nitric oxide (NO) removal rate, the nitric dioxide (NO₂) generation rate, and the overall selectivity of the photo-induced oxidation reactions. Specifically, the efficiency of NO removal was strongly influenced by the light irradiation type. Under visible light (VIS) exposure, NO removal rates ranged between 15 and 60 µg/hm², demonstrating relatively modest activity likely due to the limited energy of visible wavelengths. Conversely, under UV-A light, the removal rates increased significantly to values between 130 and 320 µg/hm². When both visible and UV-A light sources were applied concurrently, the removal rates reached their maximum, ranging from 150 to 330 µg/hm², indicating a synergistic effect of the dual-wavelength illumination, as shown in

Table 5. The photocatalytic removal rate of nitric oxides (NO), generation rate of NO₂ under investigated light conditions, and selectivity of oxidation reactions for investigated mortar samples.

Sample ID	NO Removal Rate, µg/hm2			NO2 Generation Rate, µg/hm2			Selectivity, -
	Visible Light	UV-A Light	Visible and UV-A Light	Visible Light	UV-A Light	Visible and UV-A Light	Visible Light	UV-A Light	Visible and UV-A Light
PCM-1-1	39.11	232.25	247.63	1.36	31.20	23.99	0.97	0.87	0.90
PCM-1-2	29.73	199.83	232.65	6.28	79.72	55.41	0.78	0.60	0.76
PCM-2-1	29.29	196.60	201.21	9.09	84.17	88.50	0.69	0.57	0.56
PCM-2-2	53.67	319.56	326.34	8.64	79.63	84.27	0.84	0.75	0.74
PCM-3-1	30.71	220.34	242.89	8.81	76.13	53.65	0.71	0.65	0.78
PCM-3-2	40.95	281.78	286.58	7.86	37.50	36.49	0.81	0.87	0.87
PCM-4-1	30.68	189.44	210.61	2.23	8.77	9.37	0.93	0.95	0.96
PCM-4-2	15.31	137.87	154.04	5.41	77.05	72.21	0.65	0.44	0.53
PCM-5-1	45.53	219.16	242.11	9.68	43.24	42.74	0.79	0.80	0.82
PCM-5-2	26.27	148.24	176.84	4.48	99.06	64.23	0.83	0.33	0.64
PCM-6-1	55.23	255.68	303.52	5.23	64.63	51.32	0.91	0.75	0.83
PCM-6-2	24.81	181.81	186.41	3.83	215.11	195.45	0.85	−0.08	−0.15

.

In addition to NO removal, the rate of NO₂ generation—a less desirable byproduct of incomplete photocatalytic oxidation—was also notably influenced by the illumination conditions. Under VIS light, the NO₂ generation rate was comparatively low, ranging from 1 to 10 µg/hm². However, under UV-A light alone, the NO₂ generation rate increased markedly, spanning from 8 to 216 µg/hm², reflecting the higher intensity of the oxidative reactions under UV irradiation.

Significant differences in NO₂ generation rates under UV-A light and combined UV-A + visible light were observed in some cases. For example, in the PCM-5-2 sample, the NO₂ generation rate decreased notably from 99.06 µg/hm² to 64.23 µg/hm² after adding visible light, whereas it remained approximately unchanged for the PCM-5-1 sample (43.24 µg/hm² under UV-A light vs. 42.74 µg/hm² under combined light). This phenomenon can be attributed to differences in pore structure between the samples, particularly the content of pores with diameters below 0.1 µm, which strongly affects the NO₂ generation rate, as discussed in Section 4.3. Additionally, this could be linked to the exposure of two types of photocatalysts (P25 and K7000) to light.

The PCM-5-2 sample exhibited the lowest selectivity under UV-A light, suggesting that photocatalytic reactions were less effective due to the low porosity of the cement matrix. However, as the results indicated, the inclusion of visible light with higher irradiation intensity than UV-A, and consequently greater penetration into the pore structure of the cement composite, partially mitigated this effect. This led to increased efficiency of photocatalytic reactions, which, in turn, resulted in a significantly lower NO₂ generation rate. Furthermore, this effect was likely associated with the incorporation of second-generation photocatalysts (K7000) into the cement matrix. These photocatalysts exhibit significant air purification potential under visible light, contributing to the observed behavior.

The selectivity of the photocatalytic oxidation processes also varied significantly among the different mortar samples and light regimes, with some samples exhibiting unfavorable reaction dynamics. For instance, in the case of sample PCM-6-2, the secondary photocatalytic conversion of NO₂ to nitrate (NO₃⁻) was characterized by markedly low efficiency under both UV-A and combined light conditions. This inefficiency led to a scenario where the rate of NO₂ generation exceeded that of NO removal, representing a counterproductive outcome for practical applications aimed at reducing air pollution.

An analysis of collected data and their variability was conducted and visualized via box plots (Figure 7 and Figure 8). In the case of NO removal rate in UV-A light conditions, the analyzed removal rates for all investigated samples were characterized by a total range of 181.69. The central tendency, represented by the median, was 209.50, which falls closer to the first quartile (185.63) than the third quartile (243.97), suggesting a slight skewness toward higher values. The interquartile range (the spread of the middle 50% of the data) was calculated as 58.34, indicating moderate variability within the dataset. Similar observations were made for the remaining characteristics of the photocatalytic process in the designed experiment, suggesting a moderate variability for both rates under considered irradiation conditions. Except for the outliner sample in the case of NO₂ generation and selectivity (PCM-6-2), no other outliners were identified. An increase in the variability in reaction selectivity between UV-A irradiation and UV-A + VIS was detected.

4.2. Photocatalyst Distribution

The area occupied by photocatalysts varied across all 504 analyzed EDS mappings, with most coverage concentrated in the lower ranges (up to 0.6% of the area). The highest number of mappings was observed in the 0–0.6% range (1464 grid rectangles), followed by a gradual decline in higher intervals: 0.6–1.2% (649 rectangles), 1.2–1.8% (224 rectangles), and 1.8–2.4% (84 rectangles). Beyond 2.4% of the area, the number of 52 × 50 rectangles continued to decrease, with minimal values in the 2.4–3.0% range (15 occurrences) and only 12 occurrences above 3% coverage. This distribution emphasized the dominance of areas with minimal photocatalyst coverage on the sample’s surface (Figure 9).

Analyzing spatial autocorrelation metrics (Moran’s I and Geary’s C metrics) and chi-square tests revealed varied uniformity in photocatalyst distribution across sample surfaces. The chi-square values range from 9.62 to 72.88, with consistently high p-values (~1.0), indicating that the test did not detect statistically significant spatial variability. Moran’s I and Geary’s C provided more detailed insights, revealing three distinct clustering ranges based on their values (Figure 10):

• No Clustering (Highly Dispersed Distribution of Photocatalysts): Characterized by Moran’s I values below 0.2 and Geary’s C values above 0.7, such as PCM-4-2 (Moran’s I = 0.0245, Geary’s C = 0.9639). These values indicate a highly dispersed photocatalyst distribution with minimal or no agglomeration.
• Low Clustering: Defined by Moran’s I values between 0.2 and 0.4 and Geary’s C values between 0.6 and 0.7, such as PCM-2-2 and PCM-5-1. These samples show a slight tendency for clustering, indicating a distribution with some minor aggregation but retaining partial uniformity.
• Intermediate Clustering (Localized Agglomerations of Photocatalysts): Characterized by Moran’s I values between 0.4 and the observed maximum of 0.66 and Geary’s C values below 0.6, such as PCM-6-1 (Moran’s I = 0.663, Geary’s C = 0.3373) and PCM-3-2. While not representing the highest possible clustering on the theoretical Moran’s I scale, these values indicated relatively strong clustering compared to other samples in this dataset, with notable localized agglomerations.

Intermediate clustering with localized agglomerations of photocatalysts was observed in the majority of the samples (7 out of 12). Only one sample (PCM-4-2 with the lowest overall content of TiO₂ grains over the photoactive surface among all considered samples) demonstrated highly dispersed photocatalysts on its surface, showing no signs of agglomeration. In contrast, a low level of agglomeration was evident in four of the twelve samples (Figure 11).

The average surface area covered by the Ti signal on EDS maps after processing ranged from 0.17% to 1.29%. Samples with higher TiO₂ coverage tended to exhibit stronger clustering, as evidenced by high Moran’s I and low Geary’s C values. For example, PCM-6-1, with the highest clustering metrics (Moran’s I = 0.6631, Geary’s C = 0.3373), also displayed a relatively high TiO₂ coverage of 1.13%. Similarly, PCM-3-2, with Moran’s I = 0.6186 and Geary’s C = 0.3787, demonstrated a high TiO₂ coverage of 1.13%. These findings suggest that higher TiO₂ concentrations likely promoted localized agglomerations during application, leading to more pronounced clustering.

In contrast, samples with lower TiO₂ coverage, such as PCM-4-2 (coverage = 0.17%), exhibited low clustering metrics (Moran’s I = 0.0245, Geary’s C = 0.9639), indicating a highly dispersed distribution. This supports the hypothesis that reduced photocatalyst application may result in a more uniform distribution with fewer agglomerations (Figure 12).

4.3. Pore Structure

The total porosity of the investigated photocatalytic mortars exhibited comparable values across the different series, ranging from 11.58% to 14.53%. Despite this overall similarity, the distribution of pore sizes within the mortars demonstrated significant variability depending on the sample. The total pore structure was categorized into gel pores (<10 nm), pores associated with crystallized hydration products (10–100 nm), capillary pores (100–1000 nm), and macropores (>1000 nm), with notable differences observed in their relative contributions to the total pore volume (Figure 13).

All samples displayed a relatively uniform proportion of gel pores, accounting for 7.25% to 9.56% of the total pore volume. Likewise, the content of macropores also exhibited a narrow range, contributing between 5.12% and 7.87% to the total pore volume. However, the content of capillary pores varied considerably among the samples, ranging from as low as 5.31% to as high as 24.14% of the total pore volume. Similarly, pores with diameters between 10 and 100 nm were characterized by a wide distribution, ranging from 58.49% to 80.45% of the total pore volume. These variations in pore size distribution suggest that while the overall porosity values remained consistent across mortar samples, the microstructural characteristics, particularly the balance between smaller gel pores and larger capillary and hydration product-linked pores, differed substantially.

Such differences in pore size distribution can have a significant impact on the physical and functional properties of the cementitious composites, including their photocatalytic performance, mechanical strength, durability, and other properties [31]. The variability in capillary pores and hydration-related pores is particularly relevant, as these characteristics can influence the accessibility of reactive surfaces for photocatalytic activity and the transport of water and ions within the composite.

5. Discussion

The development of sustainable and efficient composites is essential for reducing the environmental impact of the construction industry [32]. Photocatalytic cementitious composites have attracted considerable attention for their ability to improve urban air quality by facilitating the removal of gaseous pollutants, particularly nitrogen oxides (NOx) [33,34,35]. The photocatalytic process is initiated when the photocatalyst, typically TiO₂, is exposed to light with energy equal to or greater than its bandgap energy. This exposure generates electron/hole pairs (e⁻/h⁺ pairs) on the photocatalyst’s surface. The holes (h⁺) in the valence band possess strong oxidative properties, while the electrons (e⁻) in the conduction band exhibit reductive characteristics.

The holes interact with water molecules (H₂O) adsorbed on the photocatalyst’s surface to produce hydroxyl radicals (•OH), while the electrons reduce oxygen molecules (O₂) to form superoxide anions (O₂⁻•). These hydroxyl radicals and superoxide anions are highly reactive oxidative species that drive the degradation of pollutants through photocatalytic reactions. Nitric oxide (NO) molecules adsorb onto the surface of the photocatalyst, where they are oxidized by •OH radicals into nitrogen dioxide (NO₂). The NO₂ molecules then undergo further reactions with oxidative species, such as •OH and O₂⁻•, to form nitrate ions (NO₃⁻), which are less harmful and can be easily removed or washed away.

The overall photocatalytic performance of cementitious composites is determined by the interplay of numerous phenomena that collectively influence the rates of chemical reactions initiated by exposure to electromagnetic radiation of specific wavelengths [36,37,38,39,40,41,42]. A critical factor in this process is the integration and distribution of photocatalysts within the composite matrix, particularly at its surface. The effectiveness of photocatalyst embedment and the specific surface properties of the composite, including porosity, roughness, and wettability, directly impact the initiation and intensity of photocatalytic reactions [18,43]. These surface characteristics also modulate the interaction between reactive species, such as free radicals, and pollutants, thereby influencing the duration and efficiency of contact between exposed photocatalytic grains and polluted medium.

Moreover, the internal composition of the composite plays a pivotal role in the spatial distribution of nanometric photocatalytic modifiers throughout its volume. This distribution determines the uniformity of active photocatalytic sites across the surface, affecting the effective photoactive area available for pollutant degradation [43,44]. The homogeneity or heterogeneity of this distribution can either enhance or inhibit the overall photocatalytic performance. Thus, the balance between composite formulation, surface engineering, and photocatalyst integration is critical for optimizing the photocatalytic efficacy in concrete technology.

5.1. TiO₂ Distribution and Air Purification Efficiency Correlation

A significant correlation was observed between the average surface covered by TiO₂ and the NO removal rate, whereas no correlation was detected between the average TiO₂ coverage and the NO₂ generation rate. Notably, only samples with a TiO₂ coverage exceeding 1% achieved NO removal rates greater than 250 µg/hm² under UV-A + VIS conditions. Conversely, samples with a TiO₂ coverage below 1% consistently exhibited NO removal rates below this threshold. Furthermore, the majority of samples were clustered within NO removal rates of 150–250 µg/hm² and TiO₂ coverage levels of 0.2–0.6% (Figure 14).

Using the analysis of regression, three equations were derived: Equation (3) for NO removal rate under combined light conditions, Equation (4) for UV–A light, and Equation (5) for visible light alone. A natural logarithmic approximation was utilized, assuming that the NO removal rate would approach zero as the average TiO₂ coverage nears zero.

$${NORR}_{COMB}=29.92·\ln \left(ATiC\right)+255.65$$

(3)

$${NORR}_{UV–A}=27.69·\ln \left(ATiC\right)+235.16$$

(4)

$${NORR}_{VIS}=4.84·\ln \left(ATiC\right)+38.78$$

(5)

where ${NORR}_{COMB}$, ${NORR}_{UV–A}$, ${NORR}_{VIS}$ are the NO removal rates for combined light, UV–A light, and visible light, respectively, µg/hm²; $ATiC$—Average TiO₂ coverage, %.

All regression equations demonstrated a strong correlation with the experimental data. The coefficients of determination (R²) were 0.80 for model (3), 0.76 for model (4), and 0.62 for model (5). TiO₂ grains on the surface of the composite provided accessible active sites for photocatalytic reactions to occur. The distribution and density of these grains directly correlate with the available reaction sites for NO molecules, impacting the overall rate of pollutant degradation. A higher concentration of surface-exposed TiO₂ grains enhanced the likelihood of NO molecules interacting with photoactivated sites, thus increasing the reaction rate. Although the primary oxidation reaction efficiency was correlated with TiO₂ exposure to external radiation, the NO₂ generation rate was not, indicating a different phenomenon driving the observed difference in the selectivity of investigated photocatalytic reactions.

5.2. Pore Structure and Air Purification Efficiency Correlation

The analyzed samples exhibited marked variability in the volume and distribution of capillary pores (1.0–0.1 µm) near the composite surface, with pore volumes ranging from approximately 24.14% to 5.31% of the total pore network. This substantial difference in pore characteristics highlights their influence on the material’s functionality, particularly in processes occurring within the near-surface zone of the cement matrix.

When the capillary pore volume is sufficiently high, the reactants are effectively channeled to deeper photocatalytic sites, optimizing the interaction and enhancing the overall efficiency of the photocatalytic reactions. This transport mechanism supports a balanced sequence of photocatalytic reactions, including the primary oxidation of nitrogen monoxide (NO) to nitrogen dioxide (NO₂) and the subsequent secondary reactions that are crucial for pollutant removal.

A significant reduction in the volume of capillary pores imposes severe limitations on the diffusion of reactants. In such cases, the activity of photocatalytic sites becomes constrained to the outermost surface layer of the composite. This confinement disrupts the balance required for effective photocatalytic performance—specifically, the restricted transport of reactants prevents the secondary oxidation reactions needed to sustain the efficacy of the primary process, such as the further oxidation of NO₂ (Figure 15). Using the analysis of regression, two equations were derived, describing that trend in two irradiation conditions—for UV-A light (6) and for combined light sources (7).

$${NOGR}_{UV-A}=0.0026·e^{1.0087·PU01}$$

(6)

$${NOGR}_{COMB}=0.0010·e^{1.0981·PU01}$$

(7)

where ${NOGR}_{COMB}$, ${NOGR}_{UV–A}$ are the NO₂ generation rates for combined light and UV–A light, µg/hm²; $PU01$—Pores under 0.1 µm, % of total sample volume.

The overall photocatalytic efficiency is diminished, as the reduced capillary pore volume inhibits the material’s ability to maintain a consistent and effective interaction between pollutants and photoactive sites. In extreme scenarios, the markedly reduced contact time between reactive species and photocatalytic sites can disrupt the balance of the reaction kinetics, leading to a situation where the rate of nitrogen dioxide (NO₂) generation exceeds the rate of nitrogen monoxide (NO) removal (as was the case for PCM-6-2 sample). This imbalance not only undermines the efficiency of the photocatalytic process but also results in a net production of NO₂, thereby rendering the overall selectivity of the photocatalytic system negative. Such conditions underscore the critical necessity of maintaining optimal reaction environments, which are intrinsically linked to the pore network characteristics of the photocatalytic composite. The microstructural parameters of the pore network, while not directly altering the surface morphology of the composite, play a pivotal role in governing the interaction dynamics between the two involved media: the photocatalytic cement matrix and the atmospheric pollutants. By influencing the diffusion pathways, reactant accessibility, and transport mechanisms, the pore network indirectly determines the efficiency and coordination of primary and secondary photocatalytic reactions. These interactions are essential for sustaining a synergistic reaction sequence, ensuring effective pollutant degradation while mitigating the risk of imbalanced reaction rates that could compromise the composite’s overall functionality.

5.3. Predictive Modeling of Air Purification Efficiency

The conducted research identified two key phenomena that statistically significantly influence the air purification performance of cementitious composites. The first was the contribution of near-to-surface TiO₂ grains, which drive the primary photocatalytic oxidation reaction responsible for nitrogen monoxide (NO) removal. The second was the influence of the near-to-surface pore network characteristics, which affect the intensity and efficiency of the secondary reaction responsible for the removal of nitrogen dioxide (NO₂) generated during the primary oxidation process. These phenomena underscore the interplay between surface chemistry and material microstructure in determining the overall photocatalytic performance of cementitious composites.

Two predictive models were developed and compared to evaluate the NO_x removal rate in combined light and incorporate these two critical phenomena. The models employed two independent variables: (1) the percentage of the total sample volume occupied by pores under 0.1 µm in diameter, which represents a threshold below which fluid transport is significantly impeded, and (2) the area of the photoactive external surface of the composite covered by TiO₂ grains. These variables reflected the dual importance of microstructural pore characteristics and surface-active photocatalytic regions in mediating pollutant removal processes. Most studies in the field predominantly emphasize modifications to the properties of photocatalytic materials themselves, such as the particle size, surface area, or doping levels, in order to enhance their photocatalytic performance [45,46]. However, the influence of the surrounding cement matrix, which serves as the host for these materials, has been largely overlooked. In this study, the authors demonstrated that the structure of the pore network within the cementitious composite exerts a statistically significant impact on the overall air purification efficiency. This finding highlights the importance of considering not only the characteristics of the embedded photocatalytic materials but also the structural attributes of the matrix, which play a crucial role in determining the efficacy of pollutant removal. By integrating both aspects—material properties and matrix characteristics—into a unified modeling framework, this study provides a more comprehensive understanding of how these factors interact to influence the composite’s photocatalytic performance. The dependent variable modeled was the net NO_x removal rate, defined as the rate of NO removal reduced by the corresponding NO₂ generation rate.

The first model was a superposition of regression Equations (3) and (7), giving the regression Equation (8)—Figure 16.

$${NOXRR}_{comb}=NO-{NO}_2=29.92·\ln \left(ATiC\right)+255.65-0.0010·e^{1.0981·PU01}$$

(8)

where ${NOXRR}_{comb}$ is the NO_x removal rate, µg/hm², and $ATiC$ is the average TiO₂ coverage, %; $PU01$—Pores under 0.1 µm, % of total sample volume.

The adjusted coefficient of determination ($R_{adjusted}^2$) for this model was 0.89, indicating a good correlation between the estimated and experimental data. However, the Mean Absolute Percentage Error (MAPE) was 24.51%, which exceeds the 20% threshold, suggesting low predictive accuracy. As a result, a new regression equation was developed based on the experimental data, including possible interactions between two investigated independent variables.

Through stepwise regression using a second-degree polynomial function, a second model was derived, as represented by Equation (9) and illustrated in Figure 17.

$${NOXRR}_{comb}=1305.73·PU01-69.81·{PU01}^2+8.87·PU01·ATiC-5935.73$$

(9)

The adjusted coefficient of determination for this model was 0.98, with a MAPE of 8.34%. These results demonstrated an excellent fit of the estimated data to the experimental data, as well as a high predictive accuracy of that model.

This approach captured the balance between pollutant degradation and byproduct formation, offering a more comprehensive measure of air purification efficiency. The model ultimately provides a direct indicator of the composite’s ability to transform NO_x pollutants into stable end products, such as nitrates, which can be washed off the surface through precipitation.

Incorporating both identified phenomena into a unified analysis revealed a substantial dependence of the overall photocatalytic performance on the pore network characteristics of the composite, as well as the interaction between two assumed independent variables (Figure 18). This finding highlighted the critical role that microstructural attributes play in governing the balance of photocatalytic reactions. Specifically, it was observed that the composite’s photocatalytic efficiency diminished due to suboptimal pore network characteristics, particularly the impact on the nitrogen dioxide (NO₂) generation rate, which far outweighed any performance enhancements achievable through increased exposure to near-surface TiO₂ grains.

The water-to-cement (w/c) and cement-to-sand (c/s) ratios play a pivotal role in shaping the pore network characteristics, thereby significantly influencing the photocatalytic efficiency of the composite. A higher w/c ratio generally enhances capillary porosity, increasing the surface area available for photocatalytic reactions; however, excessive porosity can compromise mechanical strength and durability [43]. Conversely, a lower w/c ratio results in a denser matrix with lower capillary pores, potentially restricting the interaction of reactive species with the TiO₂ photocatalyst. Similarly, variations in the c/s ratio impact the distribution and availability of active photocatalytic sites [47]. A higher cement content (higher c/s ratio) improves photocatalyst dispersion but must be optimized to avoid workability and shrinkage challenges. The interplay between these ratios underscores the importance of carefully tuning the material composition to balance photocatalytic performance with structural and environmental considerations.

The disproportionate influence of the pore network stems from its role in modulating the transport and diffusion of reactive species, which directly affects the secondary oxidation reactions responsible for NO₂ removal. A poorly optimized pore network not only impedes the transport of these species but also confines photocatalytic activity to the surface layer, thereby increasing the likelihood of elevated NO₂ accumulation. Given that NO₂ is considerably more harmful to human health than NO, with greater respiratory toxicity and environmental impact, the implications of this imbalance are particularly significant for the design of photocatalytic composites.

Recognizing these dynamics, the authors adopted a targeted approach to mitigate the risk of producing composites characterized by simultaneously high NO removal rates and high NO₂ generation rates. The prioritization of reducing NO₂ concentrations was central to this strategy, as it aligned with the goal of air purification technologies to minimize the presence of the most hazardous pollutants. By focusing on optimizing the pore network characteristics to support effective secondary oxidation processes, this approach ensures that the composites achieve a favorable balance between NO removal and NO₂ mitigation.

The most stable and efficient photocatalytic performance was modeled for cementitious composites exhibiting specific microstructural and surface characteristics. The optimal configuration was observed in composites where pores under 0.1 µm in diameter accounted for up to 9.6% of the total sample volume, coupled with a TiO₂ coverage of at least 0.5% on the photoactive surface. Under these conditions, the NO_x removal rate surpassed 150 µg/hm², with the selectivity of the oxidation reactions achieving values greater than 0.80, indicating a high efficiency in converting NO_x into less harmful end products, such as nitrates.

The selectivity metric reflects the composite’s ability to minimize the formation of intermediate or secondary pollutants, such as NO₂, during the photocatalytic process. A high selectivity value underscores the system’s capacity to sustain a balanced and efficient reaction sequence, which is critical for both environmental and health considerations. The optimized pore network contributes to effective reactant transport and prolonged contact time with the photocatalyst, while the sufficient TiO₂ surface coverage ensures adequate photoactive sites for initiating and sustaining the reactions.

The modeling and subsequent analysis were conducted under irradiation conditions representative of real-world environmental exposure, where both visible light and UV-A light sources were active. These conditions are particularly relevant, as UV-A light constitutes approximately 1–3% of total solar energy at ground level, making it a consistent driver of photocatalytic processes in outdoor applications. By simulating these mixed-light conditions, this study provided a realistic assessment of the composite’s performance under typical solar irradiation scenarios, enhancing the practical applicability of the findings.

This analysis highlights the importance of tailoring both the microstructural and surface properties of photocatalytic composites to achieve optimal performance metrics. The identified parameters for pore size distribution and TiO₂ coverage represent a fine balance between maintaining structural integrity and maximizing photocatalytic activity. These insights can inform the design and engineering of next-generation cementitious composites aimed at air purification, ensuring their efficacy and reliability under diverse environmental conditions.

6. Conclusions

The presented comprehensive perspective on the interplay between surface photocatalyst distribution and pore network structure underscores the complexity of designing effective photocatalytic composites, highlighting the necessity of prioritizing health-oriented outcomes, such as NO₂ reduction, in the development of these composites. Based on the conducted research, several key conclusions were drawn regarding the relationship between the properties of photocatalytic cementitious composites and their photocatalytic performance conducted in a laboratory setting mimicking the pollution and irradiation conditions present in Warsaw, Poland, in autumn/winter conditions:

• An increase in the area of the photoactive surface covered with TiO₂ grains significantly increased the NO removal rate in all considered irradiation conditions;
• A reduction in the content of pores of diameters exceeding 0.1 µm (capillary pores and macropores) significantly increased the NO₂ generation rate of investigated composites in all considered light sources, impeding the quality of secondary oxidation reactions;
• A synergistic influence of both pore network characteristics and TiO₂ exposure was identified as two main phenomena driving the overall photocatalytic performance of investigated cementitious composites;
• Investigated cementitious composites were characterized by air purification properties allowing for significant reduction in the concentration of NO_x in irradiation conditions mimicking those occurring commonly in urban settings, confirming their potential to passively offset urban pollutants;
• Future research should focus on optimizing the balance between photocatalyst distribution and pore structure to minimize undesirable byproducts like NO₂ while enhancing overall pollutant removal efficiency;
• The durability and long-term performance of these composites under varying environmental conditions should be explored further to ensure sustained effectiveness in practical applications.