**1. Introduction**

Air pollution has been recognized as the single biggest environmental threat to human health, based on its notable contribution to disease burden [1]. In this sense, European air quality regulations related to particulate matter (hereinafter PM) have been established and significantly modified in the last decades. In the 1990s, only total suspended particles (TSP) were regulated. However, since Directive 1999/30/EC entered into force, limit values for PM10 (particulate matter which passes through a size selective inlet with a 50% efficiency cut-off at 10 μm aerodynamic diameter) have been established, specifically, an annual limit value of 40 μg PM10 μg/m<sup>3</sup> with a maximum of 35 days of exceedances of the daily limit value of 50 μg/m3. This PM10 can be divided into two different categories: coarse fraction, which is mainly deposited in the tracheobronchial region (2.5–10 μm), and fine

5

**Citation:** Celades, I.; Sanfelix, V.; López-Lilao, A.; Gomar, S.; Escrig, A.; Monfort, E.; Querol, X. Channeled PM10, PM2.5 and PM1 Emission Factors Associated with the Ceramic Process and Abatement Technologies. *Int. J. Environ. Res. Public Health* **2022**, *19*, 9652. https://doi.org/10.3390/ ijerph19159652

Academic Editors: Yinchang Feng and Paul B. Tchounwou

Received: 19 April 2022 Accepted: 29 July 2022 Published: 5 August 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

fraction (<2.5 μm, including ultrafine particles (<0.1 μm)), which can penetrate deep into the lungs and translocate to the other parts of the body [2]. In this regard, in the last years, epidemiologic studies [3–6] have evidenced the negative effect of fine fraction on health. For this reason, in 2008, Directive 2008/50/EC added PM2.5 fraction (particulate matter which passes through a size-selective inlet with a 50% efficiency cut-off at 2.5 μm aerodynamic diameter) and set an annual target value of 25 μg/m3.

In the same line, the recent publication in 2021 of the document WHO Global Air Quality Guidance [1] promotes the necessity to reduce the limit values because of their impact on health. In this regard, this document defines quantitative health-based recommendations for air quality, expressed as either long- or short-term concentrations of different key air pollutants. Specifically, Air Quality Annual Guidance levels of 15 μg/m<sup>3</sup> and 5 μg/m<sup>3</sup> are recommended for PM10 and PM2.5, respectively. These guidelines are not legally binding standards; however, they provide countries with an evidence-informed tool which they can use to inform legislation and policy.

The main contributors of PM are traffic, natural phenomena, combustion in agriculture, domestic fuel burning, and industry [7]. In fact, the emissions into the air are one of the main environmental impacts from industrial activities. With regard to industrial emissions, as with air quality, regulations are becoming increasingly restrictive in terms of permitted concentrations and have broadened the parameters of interest [8]. This behavior is driven following the enforcement of Industrial Emissions Directive (IED, Directive 2010/75/UE) and Integrated Pollution Prevention and Control (IPPC, Directive 1996/61/EC), where Emission Limit Values associated with Best Available Techniques (BAT-AELs) have been established according to BAT Reference Documents (BREFs). As an example, for the ceramics industry (CER-BREF [9]), a generic BAT-AELs for dust is 30–50 mg/Nm3. Nevertheless, it should be highlighted that in the BREFs updated after 2012 (cement, wood, ferrous metals, non-ferrous metals, large combustion plants, glass, and waste incineration [10–16]), the limits established for dust are becoming much more restrictive (1–20 mg/Nm3).

In fact, in the discussion and approval of the recent BREFs, PM10 and PM2.5 are included as parameters to be monitored, but when deriving BAT-AELs, the limit is only established for TSP including PM10 and PM2.5 [17]. The PM10 parameter has appeared for the first time for emissions, additionally to the usual TSP, in tools derived from the IED, knows as Pollutant Release and Transfer Register in Europe (E-PRTR) and Spain (PRTR-Spain [18]).

On the other hand, PM2.5 determination can be deemed essential not only because of its potential impact on health, but also because it allows for the detection of anthropogenic particulate pollutants, excluding crustal particulate interference [19]. However, this parameter alone does not seem to be adequate to quantify the impact of some industries with significant primary particulate emissions (such as those of ceramics [19]). For this reason, it is necessary to obtain accurate information on both fractions (PM10 and PM2.5) in order to identify the contribution of different particulate matter sources and, therefore, to establish specific measures that allow for the improvement of air quality [20].

In Spain, according to the values declared in PRTR-Spain, more than 8381.4 tons of industrial primary PM10 were emitted into the atmosphere in 2020, of which more than 10% corresponds to ceramic industries [18]. In fact, air quality studies performed in ceramic areas have evidenced the influence of ceramic industry on air quality not only by the presence of high PM concentration, but also for the significant levels of different heavy metals [21–24]. The contribution of ceramic and related industries to PM10 and heavy metals which may be considered as tracers of the ceramic industry [24], information which is available in PRTR, is shown in Figure 1. In Table 1, the main characteristics of ceramic process emissions are shown.

**Figure 1.** Contribution of ceramic industries (3g) to some air quality indicators in Spain, 2020 [18].

**Table 1.** Ceramic process emissions.


<sup>1</sup> Hot channeled refers to those medium–high-temperature processes.

As it can be drawn for Table 1, the ceramic tile production process may generate both channeled and diffuse emissions:


measures: related to reducing the use of raw materials that could contain hazardous components; and (2) secondary measures: different abatement technologies are available as wet scrubber systems, Venturi type, fabric filters, and electrostatic precipitators.

Although the impact of the ceramic industry on air quality is well known [21–24], PM emission factors related with channeled emissions from ceramic process are not available in the EMEP/EEA air pollutant emission inventory guidebook 2019 [25]. There are some reference documents [26] which are based on US-EPA (Environmental Protection Agency) documents and previous studies [27–46] in which we can check on emission factors for spray-drying, drying, glazing, and firing resulting in a global emission factor between 2.4–11.1 kg TSP/ton depending on the implemented mitigation measures. Nevertheless, these emission factors are only available for TSP. In order to extend the information in public inventories such as E-PRTR and harmonize the key control parameters between air quality and industrial emissions (TSP, PM10, and PM2.5), the rationale of the present study was, firstly, to develop a sampling methodology based on the previous study performed by Erlich et al. [47,48] and other previous studies performed at industrial scale [49–55]; secondly, to determine PM10, PM2.5, and PM1 emission factors associated with different stages that take place during the ceramic tiles manufacturing process.

To this aim, this study was performed in a wide variety of facilities located in the ceramic production area of Castelló (Figure 2). This area extends from the coastal flat (mainly occupied by residential areas and orange tree plantations) to the mountain chain of La Cruz. This area is the largest ceramic-tile-producing zone in the EU, accounting for a turnover in 2020 of approximately EUR 3842M [56] and EUR 1200M [57] for ceramic tiles and frit and pigments, respectively. As a consequence of the high concentration of ceramic and related industries in a small area, an Air Quality Plan [58] was elaborated in 2008 to implement high-efficiency PM emissions abatement technologies in ceramic facilities and to replace impurity-bearing raw materials.

**Figure 2.** Map of the Castelló ceramic cluster [59].

#### **2. Methodology**

The method to perform the PM monitoring campaigns is based on the inertial separation of the PM target fractions and its subsequent gravimetric determination. The applied methodology was performed in accordance with the reference standards [60,61] and the specific previous studies focused on ceramic emissions [62,63], and its application allowed us to obtain the chemical characterization of the ceramic PM channeled emissions.

#### *2.1. Physical Characterization of the PM Emitted*

The physical characterization allowed for the determination of particle size distribution (PSD), mass fractions (wx: w10, w2.5, and w1) and specific emission factors (EF10, EF2.5, and EF1) for each process stage. Measurement campaigns were carried out at several ceramic companies which manufacture wall and floor tiles and frits (Table 2). All measurements were carried out under real operating conditions. The sampling period was chosen in such a way that sufficient mass is collected to permit weighing with the required accuracy without overloading the stages. Since the total PM concentrations were usually low at the tested industrial plants, very long sampling times must be provided for the reasons mentioned.

**Table 2.** Industrial scenarios.


It should be highlighted that, in some stages, measures were performed before and after the cleaning system (cyclone, wet scrubber, fabric filter, and electrostatic precipitator).

With this aim, experimental measures were taken at industrial scale using a cascade impactor (Anderson Impactor type Mark III). This impactor is designed to meet the specifications reported by VDI 2066 [64,65] and to fractionate suspended particles into different sizes categories according to their inertia.

The cut-size associated with each impactor stage depends on flow and temperature of the airstream. To calculate the PSD and mass fractions of interest (w10, w2.5, and w1) easily and accurately, the results need to be adjusted to a distribution. This distribution could be the log-normal one, which is the most usual to treat PSD data because, from the mathematical point of view, it ensures that all obtained values are positive and, therefore, they have a physical meaning. From the literature review, different types of distributions has been identified, which yield very good results in this field, such as the Rosin–Rammler–Sperling– Bennet distribution (RRSB) [47,66]. For this reason, the results obtained with log-normal distribution in the present study were compared with those obtained by RRSB distribution.

The procedure applied to calculate the cumulative log-normal and RRSB distribution is described in Figure 3.

**Figure 3.** Flow chart of the mathematical treatment used to calculate PSD and wx. D: geometric diameter (μm); S: geometric deviation; R: correlation index.

The detailed assessment of the proposed methodology was carried out in previous studies [62,63]. The evaluation criteria used was the compatibility index (CI) (EN ISO/IEC 17043 [67]), which allows us to know if two results associated with their respective uncertainties are comparable. The results compared were the total concentration measurement determined by using a cascade impactor and one obtained with the standard method (EN 13284-1 [68]). The CI was satisfactory in most of the cases studied, so it was considered that there was a good correlation between the compared concentrations. Therefore, both distributions can be considered appropriate for the objectives of the present study.

Despite the favorable results obtained in this study, and following the recommendations of the standards (EN ISO 23210 [69]), the use of a cascade impactor for the quantification of the total PM concentration is not recommended in those cases where the objective of the measurement is to ensure compliance with the established regulations.

#### *2.2. Chemical Characterization*

The objective of the chemical characterization was to obtain the chemical profile of the mass fractions w10 and w2.5 from the emissions associated with the ceramic process.

The selection of the analysis technique for the determination of the mass concentration of specific elements in the particulate matter emissions of the studied industrial processes was dependent on the amount of sample required to perform the analysis. In fact, it was the main drawback of samplings carried out at industrial scale.

This situation is more critical because of the extensive implementation of Best Available Techniques in the ceramic plants, which significantly reduces the emissions and requires very long sampling times, which makes it difficult to comply with the technical criteria established in the sampling standards. In these cases, the measurements were carried on a pilot scale (emission simulator; Figure 4), whose use was evaluated in previous studies [62,63].

The PM emissions generator allows for the regulation of the flow rate and the amount of solid material fed into an airstream with an air velocity similar to industrial installations (10–15 m/s) which transfer the dust to the sampling area, obtaining a range of PM concentrations. The feasibility of this system was deemed essential to obtain enough amount of sample for chemical analysis, upholding the technical requirements of the sampling standards. In this system, the powdered material used was provided by ceramic industries from the waste captured by the fabric filters installed to abate PM emissions generated in each stage of the process (Table 2).

**Figure 4.** PM emissions generator diagram.

The sampling by means of the PM emissions generator is based on the assumption that the content of PST and fractions w10, w2.5, and w1 of the material collected by the different cleaning systems (fabric filter and electrostatic precipitator) is similar to the composition of the w10, w2.5, and w1 of the emissions generated after the abatement system. This assumption is based on the fact that the temperature of the gases as they pass through the cleaning systems is of the same order as the emission temperature, and therefore, a priori, it is not expected that condensation processes, which could modify the chemical composition of the issued PM, take place.

The device used for collecting the sample was a Tecora cyclone, designed to meet the specifications reported by USEPA in the Method 201A [70] and to measure PM10 and PM2.5 in stack emission [62,63]. This device allowed us to obtain the required amount of sample to subsequently perform the chemical analysis of the PM10 and PM2.5 captured. The cyclone is required since the cascade impactor has different stages and, therefore, it would be very difficult to obtain enough mass of sample for the subsequent analysis using this device.

The sampling of particulate matter was carried out by means of quartz glass filters (QF20 Schleicher and Schuell). Once the PM concentrations were obtained by weighting the filters using standard procedures, one-half of each of them was digested and analyzed following the method by Querol et al. [19,71]. This method is based on an acid attack using low-pressure Teflon bombs. The solution obtained was then centrifuged and analyzed by (a) inductively coupled plasma–atomic emission spectrometry (ICP-AES) for major elements, and (b) inductively coupled plasma–mass spectrometry (ICP-MS). A quarter of each filter was used to analyze boron by the Azomethine-H method. Finally, the last quarter was sometimes used for the morphological characterization.

#### *2.3. Summary of Sampling Campaigns*

In total, more than 100 measurements were performed (pilot scale and industrial scale), which led to a measurement time of 1500 h (Table 3).


**Table 3.** Sampling campaigns description.

#### **3. Results**

*3.1. Physical Characterization of the PM Emissions*

3.1.1. Determination of PSD and wx

The determination of wx was obtained from the PSD, applying the mathematical treatment described in Figure 3. In this regard, two mathematical methods were evaluated: log-normal and RRSB (Section 2.1). It can be observed that both methods are comparable in all cases (Figure 5), so the log-normal model was applied, since it is the commonly used one in the surveyed literature.

**Figure 5.** Granulometric fractions w10, w2.5, and w1 calculated by the log-normal method and RRSB method.

The average fractions w10, w2.5, and w1 obtained from the PSD are shown in Tables 4 and 5. More detailed information about the average PSD is shown in the Supplementary Material (Figures S1–S3 and Tables S1 and S2). The average process stage PSD was calculated from the sum of the mass of the particles (within the same size range) of each of the individual samplings.


**Table 4.** w10, w2.5, and w1 obtained during milling, pressing, and glazing (ambient emissions).

**Table 5.** w10, w2.5, and w1 obtained during drying, spray-drying, firing, and frits melting (mediumand high-temperature emissions).


In order to compare the results, graphs representing the cumulative probability (cumulative mass expressed in %) versus particle diameter were produced (Figures S1–S3). This type of graph is easy to interpret and yields a straight line whenever the characterized particles come from a single source or when they have similar sizes, even if they come from several sources [72–74].

The average wx (expressed in %), as a function of the particle size, was grouped attending to the following criteria, to be easily understood:


It is remarkable the long sampling times (>40 h) required to determine the wx and PSD in those stack emissions with low particle concentrations (dryers or emissions after treatment).

#### 3.1.2. Determination of EF

This section sets out the specific emission factors obtained for the PM10, PM2.5, and PM1 for the different stages of the ceramic process, expressed as mgPMx/m<sup>2</sup> or mgPMx/kg depending on the characteristics of the processed product. For this purpose, a specific weight of 21 kg of spray-dried granulate/m2 was considered and the specific flow rates were obtained from previous studies in the ceramic industry [58,59].

The sources studied were divided into two groups, based on process temperature, as was performed in Section 3.1.2, since the literature reports and results obtained from the present study evidenced that this characteristic notably influences the size and composition of the PM emitted from these sources. Emission factors are shown in Tables 6 and 7.


**Table 6.** Emission factors for milling, pressing, and glaze preparation and glazing (ambienttemperature processes).

<sup>1</sup> Specific flow rate obtained from Monfort et al., 2013 [59] and Conselleria de Medi Ambient, Aigua, Urbanisme i Habitatge, 2008 [58].

**Table 7.** Emission factors for spray-drying, drying, firing, and frits melting (medium- and hightemperature processes).


<sup>1</sup> Specific flow rate obtained from Monfort et al., 2013 [59] and Conselleria de Medi Ambient, Aigua, Urbanisme i Habitatge, 2008 [58].

#### *3.2. Chemical Composition of PM Emissions*

In this section, the average chemical profiles are shown (Table 8), including major and trace elements of PM emission from ceramic process stages. This average profile was obtained from at least three individual valid samplings for each process stage. The major elements (expressed as oxides) are those whose percentage in composition is higher than 1%, and the trace elements are those where the concentration is higher than 100 mg/kg.


**Table 8.** PM emission composition (major and trace elements) associated with different ceramic process stages.

In stages such as spray-drying, pressing, and drying, where the processed product is quite similar (in terms of chemical composition) and has not suffered any significant physical–chemical transformations (low–medium process temperature), it is considered that the chemical profile is common for the different stages. More detailed information about the identified major and trace elements can be obtained from the Supplementary Material (Figures S4–S9).

From these results (Table 8), and taking into account previous air quality studies performed in the ceramic area of Castellón, the main tracers and other legislative elements (Ni and Cd) were selected to study the segregation of these elements in fractions PM10 (Figure 6) and PM2.5 (Figure 7), associated with the different stages of ceramic process considered.

**Figure 6.** PM10 composition in the different stages of the ceramic process.

**Figure 7.** PM2.5 composition in the different stages of ceramic process.

#### **4. Discussion**

The Discussion section follows the same structure as the Results section.

#### *4.1. Physical Characterization of the Emitted PM*

#### 4.1.1. Determination of PSD and wx

Regarding the comparison between the log-normal and RRSB distributions, it can be seen that both methods are comparable since in all cases (w10, w2.5, and w1) a trend line with a slope coefficient close to 1 and a regression coefficient higher than 0.90 was obtained. In addition, none of the methods have a clear tendency to either overestimate or underestimate the results. It can therefore be concluded, considering this evaluation, that both methods can be indistinctly used, so the log-normal model was applied, since it is the most commonly used one in the surveyed literature.

In reference to the PSD obtained in the present study, the fit parameters to a log-normal distribution, obtained by applying the calculation procedure described in Figure 3 for the calculation of wx, showed good agreement (R2 > 0.90).

From the results obtained without cleaning system, wide PSD can be observed. In the case of ambient- and medium-temperature process stages, it can be due to the presence of coarse particles, in the form of granulates and agglomerates, and fine particles. The fine particles are associated with individual particles of the processed material and particles generated by the breakage of agglomerates/aggregates (Figure 8). In high-temperature processes, wide PSD is also observed as a result of different origins for PM, coarser material generated by carryover of batch particles and finer PM from volatilization–condensation processes (Figure 9).

**Figure 8.** SEM photograph of the PM emitted by the cyclones after spray-drying.

**Figure 9.** SEM photograph of the PM emitted in the frit melting stage.

Regarding the influence of the cleaning systems on the particle size of the emitted PM, it should be noted that, in most of the processes studied (except the firing stage with solid reagent injected in the exhaust stream to remove gaseous pollutants), the finest fractions are enriched after the cleaning system. Moreover, the PSDs were less wide, presumably due to the fact that the coarser fraction was highly efficiently reduced. Finally, it was considered interesting to make some specific comments on some of the stages and cleaning systems studied (Table 9).

**Table 9.** Comments about wx associated with different ceramic process stages and cleaning systems.


In general, the mean values of w10 after the fabric filters operated at high performance are high and with little dispersion (75–85%), and it is also observed that they are practically independent of the stage considered, i.e., they are not significantly dependent on the initial PSD of the stream to be treated.

In the fine fraction w2.5, the behavior is more complex (w2.5: 30–60%), probably because the main variable is not the cleaning system, but also the nature of the processed material.

#### 4.1.2. Determination of EF

In general, the use of high-efficiency cleaning systems considerably reduces the emission factors obtained for fractions w10, w2.5, and w1. Nevertheless, the type of cleaning system and other operational parameters (such as the material processed and/or process temperature) can have an influence on the obtained results:


The obtention of specific emission factors for different particle size and stage processes, including the influence of the abatement system, is considered of great practical interest for the ceramic industry, technological providers, public authorities, and research groups for performing emission inventories (e.g., E-PRTR), deriving new BAT-AELs (Emission Limit Values associated with Best Available Techniques), air pollution diagnosis studies, environmental impact studies from a lifecycle analysis perspective, and air quality assessment studies, among others.

Despite the potential use of the ceramic specific emission factors mentioned above, it is remarkable that these are not currently available in those reference emission factors guides such as the AP-42/EPA [26] and EMEP-EEA [25] and in the BAT Reference Document applicable to the ceramic and related industries (CER BREF [9], GLS BREF [12], and WGC BREF [17]).

Table 10 shows a compilation of emission factors obtained in other similar studies [47] and in the present study. In general, they are coherent PM emissions from combustion processes (e.g., firing and fusing) finer than those generated from mechanical treatments (e.g., press and milling). In one case, significant deviations were detected between comparable processes, such as isostatic pressing of minerals; this could be due to the different sizes of the material processed.

#### *4.2. Chemical Characterization of the PM Emissions*

Finally, regarding the chemical analysis, the following conclusions may be drawn:


depends on the final aesthetic requirements of the ceramic tiles produced, and this compound is not present in all glaze compositions.



**Table 10.** Compilation of emission factors.

Another aspect studied was the possible segregation of the components and elements of interest (As, Cd, Ni, PbO, ZnO, and ZrO2) in the PM10 and PM2.5 fractions for each of the process stages. It was observed that the enrichment in one or another fraction is associated with the emission mechanism of the component and/or element evaluated and with the granulometry of the source material.


• In the melting stage of ceramic frits, both As, a trace element associated with the natural raw materials that introduce boron into the composition of ceramic frits, and ZrO2, a raw material for frits and atomized granules, are enriched in the PM10 fraction, probably because the emission mechanism in both cases is mechanical in nature.

#### **5. Conclusions**

The conclusions of this study are presented in accordance with the structure followed in the previous sections.

#### *5.1. Physical Characterization*

5.1.1. Assessment of Methodology Used to Determine wx and PSD


## 5.1.2. Determination of PSD and wx


## 5.1.3. Determination of the EF


#### *5.2. Chemical Characterization of the PM Emissions*


#### **6. Future Research Lines**

From the results obtained in the present study, a series of future research lines to complement some of the results achieved are proposed:

• To develop an emission simulator in order to modify the temperature of the stream and the introduction of gases, and thus study the effect of the temperature and the composition of the gas stream on the characteristics of the particulate matter.


**Supplementary Materials:** The following supporting information can be downloaded at https: //www.mdpi.com/article/10.3390/ijerph19159652/s1: Figure S1. PSD of emissions generated in ambient temperature processes; Figure S2. PSD of emissions generated in medium-temperature processes; Figure S3. PSD of emissions generated in high-temperature processes; Figure S4. PM10 and PM2.5 composition of emissions generated during spray-drying emissions; Figure S5. PM10 and PM2.5 composition of emissions generated during pressing; Figure S6. PM10 and PM2.5 composition of emissions generated during drying; Figure S7. PM10 and PM2.5 composition of emissions generated during firing (without cleaning system); Figure S8. PM10 and PM2.5 composition of emissions generated during firing (after cleaning system); Figure S9. PM10 and PM2.5 composition of emissions generated during frit melting. Table S1: Individual samplings of ambient temperature processes; Table S2. Individual samplings of medium- and high-temperature processes.

**Author Contributions:** Conceptualization, I.C., E.M., X.Q., V.S. and S.G.; methodology, I.C., X.Q. and A.E.; software, A.E.; validation, I.C., V.S. and S.G.; formal analysis, I.C., E.M., X.Q. and V.S.; investigation, I.C.; resources, I.C. and V.S.; data curation, I.C. and A.E.; writing—original draft preparation, I.C., V.S. and A.L.-L.; writing—review and editing, I.C., V.S., X.Q. and A.L.-L.; visualization, I.C. and E.M.; supervision, I.C. and E.M.; project administration, I.C. and E.M.; funding acquisition, I.C. and E.M. All authors have read and agreed to the published version of the manuscript.

**Funding:** This study has been funded by the Ministry of Science and Technology in the framework of the National Plan for Scientific Research, Development and Technological Innovation, reference REN2003-08916-C02-01.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Acknowledgments:** The authors would like to thank IDAEA-CSIC for their committed cooperation in the sample's chemical analysis and also all the ceramic companies in which the measurements were carried out for their continuous support.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

## **References**

