**Stabilization of Municipal Solid Waste Fly Ash, Obtained by Co-Combustion with Sewage Sludge, Mixed with Bottom Ash Derived by the Same Plant**

#### **Ahmad Assi 1, Fabjola Bilo 1, Alessandra Zanoletti 1, Laura Borgese 1, Laura Eleonora Depero 1, Mario Nenci <sup>2</sup> and Elza Bontempi 1,\***


Received: 6 August 2020; Accepted: 1 September 2020; Published: 2 September 2020

**Abstract:** This study presents an innovative stabilization method of fly ash derived from co-combustion of municipal solid waste and sewage sludge. Bottom ash, obtained from the same process, is used as a stabilizing agent. The stabilization method involved the use of two other components—flue gas desulfurization residues and coal fly ash. Leaching tests were performed on stabilized samples, aged in a laboratory at different times. The results reveal the reduction of the concentrations of heavy metals, particularly Zn and Pb about two orders of magnitude lower with respect to fly ash. The immobilization of heavy metals on the solid material mainly depends on three factors—the amount of used ash, the concentrations of Zn and Pb in as-received fly ash and the pH of the solution of the final materials. The inert powder, obtained after the stabilization, is a new eco-material, that is promising to be used as filler in new sustainable composite materials.

**Keywords:** sewage sludge disposal; municipal solid waste; co-combustion; fly ash; bottom ash; heavy metal stabilization

#### **1. Introduction**

In recent years, the diffuse practice of sewage sludge (SS) land spreading has generated several concerns, due to the presence of potential contaminants (such as pathogenic agents, toxic inorganic substances, and microorganisms) [1,2], with the consequence of emerging alternatives for SS treatments strategies, such as mono- and co-combustion. For example, co-combustion has been realized with municipal solid waste incineration (MSWI). The co-combustion has several advantages, the main related to the possibility to use the already existing plants (such as MSW incineration plants), avoiding investments costs, due to the need to construct new incinerators expressly devoted to mono-combustion. Moreover, the wastes generated from co-combustion of MSW and SS, i.e., fly ash (FA), and bottom ash (BA), must be properly managed to avoid landfilling and/or pollution, generated by unsuitable treating strategies. MSWI-FA is generally considered the most problematic incineration waste, due to the presence of leachable heavy metals [3]. Several technologies for MSWI-FA stabilization have been already proposed [4–8] with the aim to promote its reuse [9–11]. Even though some recently proposed MSWI-FA treatments were defined as zero-waste technologies [12], they often require MSWI-FA pre-treatments, which need the use of some additional processes and raw materials.

In a very recently published paper, we have demonstrated that MSWI-BA can be used to stabilize MSWI-FA [13]. Indeed, after metal separation, BA is generally recovered for use in the building industry. On the contrary, FA is generally destined to landfill. However, the recently proposed strategy allows the use of a waste (BA) to stabilize another waste (FA). It is important to highlight that the proposed procedure has several advantages—it employs wastes produced at the same location, strongly suggesting the possibility to directly apply the new technology on the incinerator plant sites; in addition, it avoids the transport of wastes in different locations and the landfilling of MSWI-FA. The idea to combine different wastes to take advantage of their valuable components to reduce the contained pollutants is not new [8,14,15]. In particular, the suggestion to use wastes and by-products to minimize energy, materials, and emissions for remediation was recently defined as the Azure Chemistry approach [16]. In the present work, the results recently proposed, that involve the combination of Sewage-MSWI BA and Sewage-MSWI FA to obtain a new safe eco-material, are considered to investigate if the process can be applied also to Sewage-MSWI FA derived from co-combustion. The obtained inert can be defined as eco-materials due to the low energies and emissions required to perform the proposed stabilization [14]. In particular, the aim of this work is the investigation if different amounts of SS in the co-combustion can influence the FA stabilization procedure that must be used to reduce the leachable pollutants.

#### **2. Experimental**

#### *2.1. Materials*

In this study, FA and BA from the co-incineration of MSW and SS were recovered from the incineration plant located in Brescia (Northern Italy). In this plant, Sewage-MSWI-FA is collected with APC (air-pollution-control) residues, originating from cleaning the flue gases before emission to air. These residues consist of fine particulates, that are generally defined FA.

The co-incineration of MSW and SS was implemented in this plant since 2017. The plant normally processes an SS concentration of 7% on the flow rate of MSW. Because the aim of the present experiment is to study the effects of different SS flow rates on the co-incineration residue stabilization, the plant worked for one day, with the three separate combustion lines at different SS concentrations—0, 2, and 4 tons/h. The amount of MSW incinerated on each line was 31 tons/h.

The schema representing the operation of the first, second and third combustion lines (also reporting the SS concentrations) is shown in Figure 1.

**Figure 1.** Schema of the co-incineration experiment made on the industrial plant, where the three different combustion lines are represented.

During the experimentation day, Sewage-MSWI FA and Sewage-MSWI BA were collected from the three separated lines at predetermined time intervals of two hours (see Figure 1). At the end of the day, the ashes collected at different interval times were adequately mixed to obtain a homogeneous and representative sample of Sewage-MSWI FA and Sewage-MSWI BA for each line. In total, 6 ash samples (2 samples for each line) were recovered (see Figure 1).

The samples obtained and considered in this study can be divided into two categories—three samples of Sewage-MSWI BA obtained from the bottom of the combustion chamber and three samples of Sewage-MSWI FA obtained from the bag filters.

After sampling, Sewage-MSWI BA was manually sorted—metals and particles with diameter higher than 2 cm were separated. Then, due to the Sewage-MSWI BA moisture, a thermal treatment at 100 ◦C for about 2 h was made. After drying, Sewage-MSWI BA was grounded with Mixer Mill Retsh (MM400) and sieved until 106 μm.

Other as-received ashes are used with the aim to stabilize FA. In particular, CFA (coal combustion ash) is a by-product in thermal coal power plants. This ash is removed by a dust-collection system from the combustion gases before they are emitted into the atmosphere [17].

Flue gas desulphurization (FGD) residues are produced during the removal of sulphur oxides from coal-burning power plants. In this process, insoluble calcium sulphite and calcium sulphate solids are formed because absorbed SO2 reacts with lime in scrubbing liquor [18]. They also are a by-product of the coal combustion system. FGD residues contain high amounts of S and Ca [5]. Calcium hydroxide is very important to promote the carbonation reactions.

Both CFA and FGD residues were collected from Brescia pulverized coal thermal power plant.

#### *2.2. Stabilization Procedure*

The as-received Sewage-MSWI FA samples were stabilized following a very recent proposed technology based on the use of MSWI-BA deriving from the corresponding combustion line to stabilize FA [4]—for each FA, the procedure (defined procedure a) involved the mix of about 130 g of FA with of 20 g of BA, about 30 g of CFA, and about 40 g of FGD [4,13] (see Table 1). Another procedure, with the same amount of ashes reported for procedure a)), but that did not involve the BA addition, was also realized (defined procedure b)). Procedure b) was selected to have a comparative procedure suitable to allow us to evaluate the BA role in the stabilization. In this case, all samples exhibited a relative weight percentage, already used for similar stabilization technology [5], as follows—65% FA, 15% CFA and 20% FGD. For both stabilization procedures, all powders were carefully mixed before adding approx. 200 mL of milliQ water (Millipore DirectQ-5 TM, Millipore S.A.S., 67120, Molsheim, France); then they were additionally mixed for 20 min. The obtained eco-materials were aged in laboratory for 3 months at room temperature.



Sewage-MSWI FA: sewage and municipal solid waste incineration-fly ash; CFA: Coal Fly Ash; FGD: Flue Gas Desulphurization; Sewage-MSWI BA: Bottom Ash.

#### *2.3. Leaching Test and TXRF Analysis*

Leaching tests were carried out according to CEN EN 12457-2 regulation (EN 13055-1:2002-CN/TC 154-CEN-CEN, 2002) on all as-received powders and stabilized samples (in this case the test was made each month, after the stabilization). The procedure for the leaching test, reported by [5,19,20], consists of mixing at room temperature approximately 20 g of each sample with 200 mL of milliQ water (1:10) by means of an agitator for 2 h [20]. pH measurement by a pH-meter (Metrohm, model 827 Lab, Origgio, Italy) is conducted immediately after the leaching test on the samples filtered by 0.45 μm pore membranes.

After filtration, elemental chemical analysis is realized by Total-reflection X-ray Fluorescence (TXRF) using a S2 Picofox system from Bruker (Bruker AXS Microanalysis GmbH, Berlin, Germany) equipped with Mo tube operating at 50 kV and 750 μA and a Silicon Drift Detector (SDD). For this aim, 0.010 g of a Ga solution with a concentration of 100 mg/L is used as internal standard (Ga-ICP Standard Soluyion, Fluka, Sigma Aldrich, Saint Louis, MO, USA). It is added to the leachate solutions and homogenized by a mean of vortex shaker at 2500 rpm for 1 min. Three replicates are always prepared for each sample by adding a droplet of 10 μL of sample. By the use of a dedicated instrumental software based on mono-element profiles, the spectra are deconvolution to evaluate the peak areas. The TXRF lower detection limits (LOD) evaluated with similar experimental conditions are reported [21]. Chemical analysis of soluble elements, with atomic numbers less than 19, such Na, cannot be made by TXRF due to their low fluorescence yield [22]. Furthermore, Si cannot be evaluated, because the sample holder for TXRF analysis is made of quartz.

#### **3. Results and Discussion**

In order to understand the effects of the addition of different SS amounts to the three combustion lines, all starting ashes (FA and BA) and the obtained stabilized eco-materials were analyzed. Elemental chemical analysis using TXRF was performed to estimate the amount of leachable metals that can be found in the ashes. As reported in the experimental section, as-received FAs were analyzed as well; BA samples were pre-treated before the analysis with a reduction of the grains dimension to 106 μm. Table 2 reports the results of the leaching tests made on these ashes (BA and FA). It shows the concentration of soluble elements derived from the three combustion lines, as resulted from the TXRF analysis. Furthermore, data about BA leaching are shown.

The leaching data of the BA and FA derived from different lines are in agreement with results reported in literature, also considering the alkaline pH of the solutions [23–25]. The major leachable elements in FA are Cl, Ca, K, S and Br. Relevant quantities of other elements such as Zn, Pb and Rb are also found. The main soluble elements found in the BA solutions are Ca, Cl, K and S. The presence of a high amount of Ca in FA in comparison to BA is expected, due to the addition of lime as a stabilizing agent. Indeed, during the incineration process, acid gases are produced (like HCl and SO2) and lime is normally added to absorb these gases. A higher amount of Pb and Zn are present in FA, in comparison to BA, due to the fact that these metals are moderately volatile elements, then they are generally found in ashes collected at the chimney after a combustion process carried out at a maximum temperature about 1000 ◦C [15]. Comparing the leaching data of different combustion lines, it is very interesting to notice that in FA, the concentration of leachable Zn is substantially irrespective of the amount of co-incinerated SS. On the contrary, the amount of soluble Pb increases with the amount of SS, probably due to the presence of this metal in sewage sludge [26]. Concerning P, an element that is well-known to be abundant in SS [27], its concentration appears to decrease in FA with the increase of co-incinerated SS. On the contrary, in BA the leachable P shows an inverse behavior. These results are substantially due to the presence of this element in soluble phases (leaching tests only allow to detect soluble elements). Moreover, it means that the increase of SS amount in co-incineration seems to increase the P leachability in BA (see Table 2). Similar consideration can be extended to Cu, for BA.


**Table 2.** Results of the Total-reflection X-ray Fluorescence (TXRF) analysis and pH values of leachate solutions of Sewage-MSWI FA and Sewage-MSWI BA derived from different combustion lines.

Values are expressed as the average ± standard deviation of three TXRF measurements. Relative sensitivities for elements with \* were calculated based on a calibration curve. LOD-lower detection limit.

After the stabilization procedure, a significant decrease in heavy metal leachability can be noticed, if samples are analyzed one, two, and three months after the stabilization (see Tables 3 and 4), in comparison to data reported in Table 2. In particular, three months after the stabilization the concentration of Pb is sometimes lower than LOD by TXRF spectrometry, which is 0.002 mg/L, thus demonstrates that the stabilization procedure was effective in reducing the solubility of heavy metals and that the stabilization efficacy increases with time. Moreover, comparing data reported in Tables 3 and 4 (considering the corresponding month of stabilization), it seems that the use of BA is not so fundamental in reducing the heavy metal mobility. Indeed, eco-materials obtained applying the procedures a) and b) (the procedure b) is made without the addition of BA) sometimes show comparable Pb concentrations in their leaching solutions. Another difference concerns the Cu concentration, that is higher in samples treated by BA addition. Even if this metal appears to be stabilized after aging, it is evident that this origin can be attributed to BA due to its higher concentrations in the BA than FA (see Table 2).

In a very recent paper, the stabilization mechanism was proposed and discussed—it was shown that dissolved amorphous silica and alumina (derived from BA) in the presence of calcium ions (and in a highly alkaline environment) promote a pozzolanic reaction with FA, with the formation of cementitious compounds such as C–S–H and calcium aluminate hydrates (C-A-H). Furthermore, carbonation reactions occurred, due to the calcium hydroxide that is present in the used wastes (for example in FGD residues) [28,29].

In particular, considering the pH of all raw FA and final stabilized eco-materials, it means that all FAs have a starting pH about 12 (see Table 2). Instead, stabilized materials have a pH of about 11 (see Table 3) and 8 (see Table 4), depending on the procedure used for stabilization. Indeed, carbonation produces a reduction of pH [30].


**Table 3.** Results of the TXRF analysis and pH values of samples stabilized following the a) procedure with Sewage-MSWI FA from different lines after (1), (2) and

68

detection limit.


**Table 4.** Results of the TXRF analysis and pH values of samples stabilized following the b) procedure with Sewage-MSWI FA from different lines after (1), (2) and

69

detection limit.

Both pozzolanic and carbonation reactions have been demonstrated to be effective in heavy metal mobilization [31], but the comparison between procedures a) and b) highlights the fundamental contribution of carbonation. As explained in the introduction part, the difference in the amount of SS in the co-combustion with MSW is considered to evaluate if the addition of this waste plays a role in the stabilization mechanism. For this aim, the procedure not involving the use of BA (procedure b) is considered. Moreover, this procedure allows for the reaching of lower pH values of the obtained eco-materials (see Table 4), increasing the efficacy of stabilization.

Figure 2 reports the concentration of Pb and Zn in the leaching solutions of stabilized samples, considered after different times, versus the pH of the solution (for both procedures a) and b)). Zn and Pb are found in a high concentration in the raw FA (ranging from 9–11.7 and 92–127.1 mg/L, respectively) with high pH values (about 12), while in the stabilized samples the concentration of these elements is found to be often two orders of magnitude lower. Figure 2 clearly highlights that an increase of the aging time corresponds to a decrease of pH of the solutions [32], in agreement with the already reported evidence of carbonation. This also corresponds to a decrease of the heavy metal leachability, as already observed and discussed [4,33]. In particular, concerning Pb, three months after stabilization it is sometimes lower than LOD by TXRF spectrometry, as reported in Tables 3 and 4, therefore it cannot be found in Figure 2.

**Figure 2.** Concentration values of Pb and Zn in the leaching solution of stabilized samples (involving both procedures a) and b)) during the first three months versus the pH.

To better highlight the pH role, the variation of the concentration of Pb and Zn in the leaching solutions of stabilized samples can be considered analyzing data reported in Figure 3. In this Figure, the concentration of Pb and Zn in the solutions of stabilized samples (considering both procedures a) and b)), are plotted versus their values in the leaching solution of raw FA (initial metals concentrations, before the stabilization). Figure 3 allows for the comparison of the results of stabilization, considering that raw FA (corresponding to different combustion lines) had a different amount of leachable Pb and Zn. In particular, it is evident that all samples show a reduction of the Pb and Zn concentration, after the stabilization. Obviously, samples that contained a lower amount of leachable heavy metals in raw FA show a lower concentration of corresponding metals in the solutions of stabilized eco-materials. This means that samples with higher initial concentrations of Pb and Zn need more time (or more stabilizing agents) to reduce the leachable elements concentration.

**Figure 3.** Concentration values of Pb and Zn in the leaching solution of stabilized samples (involving both procedures a) and b)) during the first three months versus the concentration of Pb and Zn in MSWI FA.

The role of BA in the stabilization can be highlighted by analyzing data reported in Figure 4, that show the concentration of Pb and Zn during the first three months evaluated in the leaching solutions of all stabilized samples (considering both procedures a) and b)), versus the ratio between the total amount of ash (FA, CFA, FGD, and BA) and the sum of CFA and BA amount. It means that the samples that contain the higher CFA + BA amount are better stabilized in comparison to those containing a lower amount of these stabilization agents (the concentration of Pb and Zn are lower). Indeed, it is important to highlight that BA contains amorphous silica [4] and CFA contains aluminosilicate glass [13,34]. Then, it is possible to suppose that dissolved amorphous silica in the presence of calcium ions (and in a highly alkaline environment) promotes a pozzolanic reaction with FA. Then, it is possible to conclude that BA plays a fundamental role in reducing the Pb and Zn presence in solution of the produced eco-materials. Obviously, it was also shown that the pH role is fundamental, and it must be controlled and possibly adjusted to obtain the best results in terms of stability of the obtained eco-materials.

**Figure 4.** Concentration values of Pb and Zn in the leaching of stabilized samples (involving procedure a)) during the first three months versus the partition of total amount of ash and the sum of CFA and BA quantities.

The obtained stabilized eco-material is a powder that is very similar to the inert material obtained by using similar treatments but using other by-products (such as silica fume) instead of BA for stabilization. This allows for us to suppose that the obtained eco-materials may be used in some applications, already explored for similar products, as a substitute of natural resources [10,33,35,36]. In particular, it was shown that the obtained eco-materials are biologically safe [37–39], opening the interesting opportunity of the investigation of their aquatic toxicity in the next future.

#### **4. Conclusions**

The present paper concerns the study of a method for the stabilization of heavy metals contained in FAs derived from the co-combustion of MSW and SS. BA, a residue of the same process, is used as a stabilizing agent. Leaching test reveals the reduction of heavy metals in the stabilized samples, that increases with the aging (to three months). In particular, the concentration of leachable Zn and Pb in the as-received FAs decreased to two orders of magnitude in the solution of stabilized samples. In addition, the reduction of pH of the same solutions confirms the occurring of carbonation reactions, that also contributed to reduce the heavy metal leachability.

In summary, these results show that the efficacy of the stabilization procedure depends on several factors—the pH is fundamental to reduce the heavy metal leachability. In addition, the amount of leachable Pb and Zn in as-received FA is an important parameter to consider obtaining the better results—as expected, ashes containing higher quantities of contaminants require higher quantities of stabilizing agents (or more time for the stabilization). Indeed, an increase in the sum of the concentration of CFA and BA corresponds to a better result in terms of metals reduced mobility.

The obtained stabilized eco-material may be used in different applications as a substitute of natural resources.

**Author Contributions:** Conceptualization, E.B. and M.N.; methodology, E.B.; software, A.A.; formal analysis, F.B., A.Z. and L.B.; investigation, A.A., A.Z. and F.B.; data curation, F.B. and A.Z.; writing—original draft preparation, A.A. and E.B.; writing—review and editing, L.E.D., E.B., F.B., L.B. and M.N.; supervision, E.B.; project administration, E.B.; funding acquisition, E.B. and M.N. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was partially supported by FANGHI project, financed by Regione Lombardia, in the frame of the call HUB Ricerca e Innovazione.

**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**


© 2020 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 (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Microalgal Growth in Paper Industry E**ffl**uent: Coupling Biomass Production with Nutrients Removal**

**Bruna Porto 1,2, Ana L. Gonçalves 3,\*, Ana F. Esteves 3, Selene M. A. Guelli Ulson de Souza 1, Antônio A. Ulson de Souza 1, Vítor J. P. Vilar <sup>2</sup> and José C. M. Pires <sup>3</sup>**


Received: 31 March 2020; Accepted: 24 April 2020; Published: 26 April 2020

**Abstract:** Paper and pulp industries produce effluents with high phosphorus concentrations, which need to be treated before their discharge in watercourses. The use of microalgae for this purpose has attracted the attention of researchers because: (i) microalgae can assimilate phosphorus (one of the main nutrients for their growth); and (ii) growing on effluents can significantly reduce the costs and environmental impact of microalgal biomass production. This study evaluated the growth and ability of *Chlorella vulgaris* to remove the phosphorus from a secondary-treated effluent of a Portuguese paper company. Batch experiments were performed for 11 days using different dilutions of the effluent to evaluate its inhibitory effect on microalgae. Results showed that the non-diluted effluent inhibited microalgal growth, indicating that this bioremediation process is possible after a previous dilution of the effluent. Regarding phosphorus removal, promising results were achieved, especially in the experiments conducted with the most diluted effluent: removal efficiencies obtained in these conditions were (54 ± 1)%. Another interesting finding of this study was microalgal growth in flakes' form (mainly due to the compounds present in the effluent and to the pH values achieved), which can be an important economic advantage for biomass recovery after the remediation step.

**Keywords:** biomass production; *Chlorella vulgaris*; microalgae; nutrients removal; paper industry effluent; effluent treatment

#### **1. Introduction**

Paper and pulp industries require large amounts of water during their manufacturing stages. For example, the production of 1 kg of paper requires 10 to 50 L of water [1]. At the same time, large amounts of effluents (about 2000 m<sup>3</sup> d−1) are generated, presenting as main features [1,2]: (i) high chemical oxygen demand (COD, 1000–13,000 mgO2 L<sup>−</sup>1); (ii) high total suspended solids contents; (iii) non-biodegradable organic materials; (iv) adsorbable organic halogens (AOX): (v) color; (vi) phenolic compounds; (vii) high total phosphorus contents; and (viii) limiting nitrogen concentrations. Due to the large volumes involved and respective compositions, discharge of these effluents without any treatment can cause several environmental problems [1,3]: (i) colored effluents can affect aesthetics, water transparency and gas solubility in water bodies; (ii) increase in the concentration of toxic

compounds, which can affect aquatic flora and fauna; and (iii) eutrophication with consequent decrease of dissolved oxygen concentration and pH oscillations, which can negatively impact aquatic ecosystems. Therefore, treatment of these effluents is necessary before their discharge.

Among the contaminants present in these effluents, phosphorus is of particular concern, as it subsists in the effluents after the secondary treatment step and is one of the main contributors to the eutrophication phenomenon [4]. Currently applied methods to reduce phosphorus concentration in these effluents include physicochemical methods, such as precipitation using aluminum and iron salts. However, these techniques tend to be costly and to produce large amounts of sludge contaminated with the referred chemical compounds, requiring further treatment [5,6]. Therefore, microalgal cultures have appeared as a feasible alternative to conventional physicochemical methods. These microorganisms have shown their ability to effectively remove color, nutrients, such as nitrogen and phosphorus, trace metals and other compounds from the culture medium [7,8].

Microalgae are fast-growing photosynthetic microorganisms that have gained much attention in recent decades, due to their high potential in a wide variety of applications. During photosynthesis, microalgae uptake CO2 from the atmosphere or flue gas emissions, contributing to the reduction of the atmospheric concentration of this greenhouse gas [9]. These microorganisms also require inorganic sources of nitrogen and phosphorus as macronutrients, enabling the use of microalgal cultures as a tertiary treatment stage (when significant concentrations of these nutrients persist after previous treatment processes) [10]. Finally, microalgal biomass presents a very rich composition in polysaccharides, lipids, proteins, vitamins, and other valuable compounds, which make microalgae a valuable resource for several applications [11,12], such as the production of natural colorants or dyes, bioenergy, and biofertilizers. Also, effluent treatment with microalgae has the following advantages [10]: (i) reduction of nitrogen and phosphorus concentrations to levels below the legislated limits for effluent discharge (EU Directives 1991/271/EEC and 1998/15/EC); (ii) recovery/recycle of these nutrients, which production presents negative environmental impacts; (iii) increase of the oxygen concentration in the treated effluent; (iv) production of biomass that can be integrated into the value chain of the company; and (v) reduction of net carbon dioxide emissions.

Despite the need to search for eco-friendly and cost-effective remediation strategies, only a few studies have reported the treatment of pulp and paper industry effluents using microalgae [1]. Tarlan et al. [7] evaluated the removal of color, AOX, and COD from an effluent resulting from a wood-based pulp and paper Turkish company using a mixed microalgal culture (composed by *Chlorella* and diatoms). Initial composition of this effluent in terms of color, AOX, and COD was: 4018 Pt-Co, 46.3 mg L−<sup>1</sup> and 1248 mg L<sup>−</sup>1, respectively. Operating in batch mode and using three different dilutions of this effluent, resulting from the process of pulp production using red pine, the authors reported removal efficiencies of 84%, 80% and 58% for color, AOX, and COD, respectively. Gentili [13] aimed to evaluate the growth of microalgae on mixtures of municipal, dairy, and pulp and paper effluents to achieve the dual goal of nutrients removal and lipids production. The use of mixtures of pulp and paper industry effluents with municipal and dairy ones was to evaluate if these mixtures could (i) promote microalgal growth without previous dilution with freshwater; and (ii) provide the required nutrients for biomass production without the need for nutrients supplementation. Characterization of these mixtures revealed an ammonium–nitrogen (NH4–N) concentration ranging from 14.75 mgN L−<sup>1</sup> to 22.35 mgN L<sup>−</sup>1, a nitrate–nitrogen (NO3–N) concentration between 1.6 mgN L−<sup>1</sup> and 10.1 mgN L−<sup>1</sup> and a phosphate–phosphorus (PO4–P) concentration ranging between 1.06 mgP L−<sup>1</sup> and 1.25 mgP L<sup>−</sup>1. With this study, carried out in laboratory tubes, in batch mode, the authors demonstrated that the microalgae *Scenedesmus* sp., *Scenedesmus dimorphus* and *Selenastrum minutum* were able to achieve nitrogen and phosphorus removal efficiencies of 96%–99% and 91%–99%, respectively. Finally, in the study performed by Usha, et al. [1], a mixed microalgal culture (composed by two *Scenedesmus* species) was grown in different dilutions (0%–95%) of a pulp and paper mill effluent, resulting from an Indian company, with the following composition: (i) 9.932 mgN L−<sup>1</sup> of NO3–N; (ii) 30.25 mgP L−<sup>1</sup> of PO4–P; (iii) 3000.15 mg L<sup>−</sup><sup>1</sup> of COD; and (iv) 2944 mg L−<sup>1</sup> of biochemical oxygen demand (BOD). The

experiments, aimed at evaluating both biomass production and nutrients removal efficiencies, were performed in batch mode, for 28 days, using open ponds as cultivation system (outdoor conditions). Regarding nutrients uptake, the most promising results were obtained for the 40% dilution: (i) 65% for NO3–N removal; (ii) 81.3% for PO4–P; and (iii) 75% for COD; and (iv) 82% for BOD.

The main goal of the present study was to evaluate biomass production and phosphorus removal from a secondary-treated effluent of a Portuguese paper company using the microalga *Chlorella vulgaris*. Different dilutions were performed to evaluate possible inhibitory effects of the effluent on microalgal growth and phosphorus uptake ability.

#### **2. Materials and Methods**

#### *2.1. Microalgae Strain and Maintenance Medium*

The microalga *C. vulgaris* (CCAP 211/11B) was obtained from the Culture Collection of Algae and Protozoa, United Kingdom. The strain was maintained on modified Organization for Economic Co-operation and Development (OECD) culture medium [14], with the following composition (mg L−1): 119 KNO3; 12 MgCl2·2H2O; 18 CaCl2·2H2O; 15 MgSO4·7H2O; 20 KH2PO4; 0.08 FeCl3·6H2O; 0.1 Na2EDTA·2H2O; 0.185 H3BO3; 0.415MnCl2·4H2O; 0.003 ZnCl2; 0.0015 CoCl2·6H2O; 0.00001 CuCl2·2H2O; 0.007 Na2MoO4·2H2O and 100 Na2CO3.

#### *2.2. Paper Industry E*ffl*uent and Culture Conditions*

Effluent from a Portuguese paper company, collected after the secondary treatment step, was characterized (Table 1) and employed as a culture medium for microalgal growth. The methodology adopted for effluent characterization was the following: (i) COD and turbidity were determined according to the Standard Methods for the Examination of Water and Wastewater [15] (through the 5220-D and 2130-B tests, respectively); (ii) total dissolved carbon (TDC), dissolved organic carbon (DOC) and inorganic carbon (DIC) were determined using an organic carbon analyzer (TOC-VCSN, Shimadzu); and (iii) chlorides, sulfates, nitrates, nitrites and phosphates were determined through ion chromatography (ICS-2100, Dionex). Due to the low concentration of nitrogen in the effluent, when compared with typical nutritional requirements of microalgae, the effluent was supplemented with NaNO3 to achieve N:P molar ratios ranging between 6:1 and 9:1. Ratios between 5:1 and 30:1 have been considered adequate for several microalgal species [16,17].


**Table 1.** Physicochemical characterization of the paper industry effluent used in this study.

<sup>a</sup> Nephelometric turbidity unit.

Batch experiments were performed in 1—L borosilicate glass flasks (VWR, Portugal) with a working volume of 950 mL for 11 days. The raw effluent (assay 1) and four different dilutions with freshwater (assays 2–5) were used as the culture medium for microalgal growth, with nitrogen concentrations (corresponding to the sum between nitrate– and nitrite–nitrogen) ranging between 12.7 mgN L−<sup>1</sup> and 34.2 mgN L−<sup>1</sup> and phosphorus concentrations (phosphate–phosphorus) ranging between 4.01 mgP L−<sup>1</sup> and 12.3 mgP L−1. The medium was inoculated with 250 mL of *C. vulgaris* inoculum to obtain an initial biomass concentration of ~ 68 mgdw L<sup>−</sup>1. The cultures were continuously exposed to: (i) photosynthetically active radiation between 30–40 μmol m−<sup>2</sup> s<sup>−</sup>1, using a 34-W white led panel; and (ii) atmospheric air filtered with 0.45–μm nylon membranes (Specanalitica, Portugal), injected at ~ 90 L h<sup>−</sup>1, using Trixie AP 180 air pumps (Trixie, Tarp, Germany). The experimental setup is shown in Figure 1. Two independent experiments were performed for each assay.

**Figure 1.** Schematic representation of the experimental setup.

#### *2.3. Microalgal Growth Monitoring and Kinetic Growth Parameters*

Operational parameters, such as pH and temperature, were daily monitored using a SympHony SB90M5 pH-meter (VWR, Portugal). Microalgal growth was also daily assessed through optical density measurements at 680 nm (OD680) using a UV-6300 PC spectrophotometer (VWR, United States). To eliminate the interference of the effluent color on OD680 measurements, the cells were separated from the culture medium by centrifugation (at 4000 rpm, for 10 min), the supernatant was discarded, and the cells were resuspended in an equal volume of distilled water, as described by Hodaifa, et al. [18]. This procedure was repeated twice. The relationship between OD680 and biomass concentration (*X*, mgdw L<sup>−</sup>1) for *C. vulgaris* was previously established by linear regression, according to Equation (1):

$$\text{X} = 0.0024 \times \text{OD}\_{680} + 0.0030 \,\text{R}^2 = 0.9999,\tag{1}$$

Biomass concentrations were used to determine the kinetic growth parameters: (i) specific growth rates (μ, d<sup>−</sup>1); and (ii) maximum and average biomass productivities (*Pmax* and *Paver*, mgdw L−<sup>1</sup> d<sup>−</sup>1). The specific growth rates were determined from the first-order kinetic model, according to Equation (2):

$$\frac{d\mathbb{X}}{dt} = \mu\mathbb{X} \leftrightarrow \mu = \frac{\ln\mathbb{X}\_1 - \ln\mathbb{X}\_0}{\mathfrak{t}\_1 - \mathfrak{t}\_0},\tag{2}$$

where *X1* and *X*<sup>0</sup> correspond, respectively, to biomass concentration (mgdw L<sup>−</sup>1) in the end (*t*1, d) and beginning (*t0*, d) of the exponential growth phase. Biomass productivities (*P*, mgdw L−<sup>1</sup> d−1) were calculated for each pair of consecutive points, through Equation (3):

$$P = \frac{\chi\_{\mathbf{z}+1} - \chi\_{\mathbf{z}}}{\mathbf{t}\_{\mathbf{z}+1} - \mathbf{t}\_{\mathbf{z}}},\tag{3}$$

where *Xz* represents the biomass concentration (mgdw L<sup>−</sup>1) at time *tz* (d) and *Xz*<sup>+</sup>*<sup>1</sup>* corresponds to the biomass concentration (mgdw L<sup>−</sup>1) at time *tz*<sup>+</sup>*<sup>1</sup>* (d). The maximum productivity was determined from the maximum value obtained from Equation (3). On the other hand, average biomass productivities were determined according to Equation (4):

$$P\_{\text{aver}} = \frac{\chi\_{\text{f}} - \chi\_{\text{i}}}{\mathbf{t}\_{\text{f}} - \mathbf{t}\_{\text{i}}},\tag{4}$$

where *Xf* and *Xi* correspond, respectively, to biomass concentration (mgdw L<sup>−</sup>1) in the end (*tf*, d) and beginning (*ti*, d) of the cultivation period.

#### *2.4. Nutrients Removal*

Nutrients removal was evaluated in terms of nitrogen and (N) phosphorus (P) present in the culture medium/effluent. Nitrogen was assessed in the forms of nitrate and nitrite ions, whereas phosphorus was monitored through the presence of phosphate ions. From each assay, 5 mL of the microalgal suspension were periodically collected (days 0, 1, 2, 4, 7, 9, and 11). These samples were centrifuged at 4000 rpm, for 10 min, and the supernatants were filtered through 0.45–μm nylon membranes (Specanalitica, Portugal). Nitrate, nitrite and phosphate concentrations were determined in an ion chromatograph (ICS-2100, Dionex) equipped with an anion analytical column (4x 250 mm, AS11-HC) and a self-regeneration suppressor (4 mm, AERS 500). The values obtained in the first and last day of culturing were used to calculate the following removal parameters: (i) removal efficiencies (*%R*, %); (ii) average removal rates (*RR*, mg L−<sup>1</sup> d<sup>−</sup>1); and (iii) mass removal (*R*, mg L<sup>−</sup>1), as shown in Equations (5), (6) and (7), respectively:

$$\% \text{R} = \frac{\text{S}\_{\text{f}} - \text{S}\_{\text{i}}}{\text{S}\_{\text{i}}} \times 100,\tag{5}$$

$$\text{RR} = \frac{\text{S}\_{\text{f}} - \text{S}\_{\text{i}}}{\text{t}\_{\text{f}} - \text{t}\_{\text{i}}} \text{.} \tag{6}$$

$$\mathbf{R} = \mathbf{S}\_{\mathbf{f}} - \mathbf{S}\_{\mathbf{i}\nu} \tag{7}$$

where *Sf* and *Si* correspond to the nitrogen (nitrate + nitrite) or phosphorus (phosphate) concentration (mg L<sup>−</sup>1) in the end (*tf*, d) and beginning (*ti*, d) of the cultivation period, respectively.

#### *2.5. Statistical Analysis*

Each parameter shown in the present paper was expressed as the mean and standard deviation. The Tukey statistical test was used to investigate if the differences between the different effluent concentrations studied could be considered significant. These statistical tests were performed using Statistica 8.0 (StatSoft Inc., USA) and were carried out at a significance level (*p*) of 0.05.

#### **3. Results and Discussion**

#### *3.1. Microalgal Growth*

The *C. vulgaris* growth curves in raw and diluted paper industry effluent are shown in Figure 2. These results evidence the inexistence of an adaptation phase for all assays and an exponential growth

phase that lasted approximately four days. In addition, no cell decay was observed during the 11-day batch culture, indicating that the experiments could be extended for a longer period. The increase of biomass concentration during the cultivation period, as well as the lack of an adaptation phase, shows that *C. vulgaris* was able to grow in this effluent. However, biomass concentrations achieved in non-diluted effluent (assay 1) were statistically lower (*p* < 0.05) than those achieved in more diluted effluents from assays 3–5.

**Figure 2.** *C. vulgaris* cultures growth curves in raw and diluted secondary-treated paper industry effluent: (**A**) Assay 11 ; (**B**) Assay 2 ;; (**C**) Assay 3 ;; (**D**) Assay 4 ; and (**E**) Assay 5 . Error bars correspond to the standard deviation of the mean obtained from two independent experiments.

To complement the analysis from growth curves, microalgal growth parameters, such as specific growth rate, maximum biomass concentration, and maximum and average biomass productivities, were determined and presented in Table 2. From these data, it is possible to see a general increase in growth parameters from assay 1 to assay 5, i.e., from the non-diluted effluent to the more diluted one. Regarding specific growth rates, values ranged from (0.093 <sup>±</sup> 0.007) d−<sup>1</sup> to (0.16 <sup>±</sup> 0.02) d−<sup>1</sup> in assays 1 and 5, respectively. The highest values of maximum biomass concentrations were also obtained in more diluted effluents from assays 4 and 5: (249 <sup>±</sup> 14) mgdw <sup>L</sup>−<sup>1</sup> and (231 <sup>±</sup> 31) mgdw <sup>L</sup><sup>−</sup>1, respectively. Similar behavior was observed for both maximum and average biomass productivities. Maximum biomass productivities/average biomass productivities obtained in assays 4 and 5 were (30 <sup>±</sup> 3)/(16 <sup>±</sup> 1) mgdw L−<sup>1</sup> d−<sup>1</sup> and (30 <sup>±</sup> 6)/(15 <sup>±</sup> 3) mgdw L−<sup>1</sup> d−1, respectively. In opposition, maximum and average biomass productivities obtained in assay 1 were (9.8 <sup>±</sup> 0.2) mgdw <sup>L</sup>−<sup>1</sup> <sup>d</sup>−<sup>1</sup> and (6.2 <sup>±</sup> 0.1) mgdw <sup>L</sup>−<sup>1</sup> <sup>d</sup><sup>−</sup>1, respectively.

**Table 2.** Specific growth rates (μ, in d<sup>−</sup>1), maximum biomass concentrations (*Xmax*, in mgdw L<sup>−</sup>1), and maximum and average biomass productivities (*Pmax* and *Paver*, in mgdw L−<sup>1</sup> d<sup>−</sup>1) determined for *C. vulgaris* grown in raw and diluted secondary-treated paper industry effluent.


Values are presented as the mean ± standard deviation obtained from two independent experiments. Within the same column, mean values sharing at least one common letter (in superscript) are not statistically different (*p* > 0.05).

In contrast to what was observed by Gentili [13], the increment in nitrogen and phosphorus concentration did not contribute to an increase in kinetic growth parameters. Accordingly, these results may indicate inhibitory effects of the effluent on microalgae, which can influence microalgal cultures in different ways [19–21]: (i) the effluent color may act as a barrier to light penetration, thus limiting microalgal access to light and photosynthetic activity; and (ii) paper industry effluents are characterized by the presence of lignin, humic acids, furans, and dioxins and by high levels of aluminum and manganese, which exhibit toxic effects on microalgae.

Most studies regarding the bioremediation of paper industry effluents with microalgae focus on the removal of contaminants and only a few report biomass production yields. Polishchuk, et al. [20] reported that the maximum specific growth rate obtained for *Nannochloropsis oculata* grown in effluents resulting from pulp and paper industry was 0.405 d<sup>−</sup>1. Tao, et al. [19] revealed that maximum biomass concentrations achieved by *Scenedesmus acuminatus* and *C. vulgaris* grown in paper industry effluents were 291 mg L−<sup>1</sup> and 822 mg L−1, respectively. Considering the values referred in the literature, microalgal growth parameters obtained in this study were significantly lower, which can be attributed to the inhibitory effects promoted by the effluent used (in assays 1–3) and to the low concentration of some essential nutrients (in more diluted effluents of assays 4 and 5). Another explanation for the low biomass concentrations and productivities achieved may be related to the phenomenon of flakes formation observed within the cultivation period (autoflocculation). Cells' agglomeration can affect the accurate measurement of OD680 and, on the other hand, it can reduce light absorption efficiency by cells incorporated within flakes, thus resulting in lower photosynthetic activity. In this study, this phenomenon occurred due to the increase of culture pH (from 7.8 to 8.6) or due to the presence of certain compounds in the effluent, which can induce a change in the surface charge of the cells and affect suspensions' stability [22]. Despite the low microalgal growth rates, the flakes formation enables a cost-effective biomass removal after effluent remediation. The density similar to water and small size of microalgal cells difficult the harvesting process and make this step one of the most expensive within microalgal biomass production processes [22,23]. However, when cells agglomerate, an increase in

their density and size is observed, contributing to higher settling rates and allowing biomass recovery using the least expensive harvesting method: sedimentation.

#### *3.2. Nutrients Removal*

In this study, nitrogen (in the forms of nitrate and nitrite) and phosphorus (in the form of phosphate) concentrations were monitored within the cultivation time to evaluate the potential of *C. vulgaris* to uptake these nutrients from a paper industry effluent with different concentrations of both nitrogen and phosphorus. Figure 3 shows the variation of nitrogen and phosphorus concentration in each assay. Regarding nitrogen removal (Figure 3A), this element was readily assimilated by *C. vulgaris* in the diluted effluents (assays 2–5). In the raw effluent (corresponding to assay 1), a two-day delay was observed in nitrogen assimilation, which may be related to the adaptation of the microalga to these conditions. Regarding the assimilation patterns observed in assays 2–5, these were approximately linear for assays 2–4, with nitrogen concentration decreasing gradually during the cultivation time. On the other hand, in assay 5, corresponding to the more diluted effluent experiments, nitrogen concentration decreased until the seventh day of culturing and then it was maintained approximately constant. This behavior may be attributed to a decrease in photosynthetic activity, as nitrogen concentration decreased, and explains the lower biomass concentrations achieved in assay 5 when compared to the one obtained in assay 4 (according to Table 2, (231 <sup>±</sup> 31) mgdw <sup>L</sup>−<sup>1</sup> and (249 <sup>±</sup> 14) mgdw L−1, respectively). Also, at the end of the cultivation time, nitrogen concentration remaining in cultures corresponding to assays 4 and 5 was approximately the same ((2.81 <sup>±</sup> 0.05) mgN L−<sup>1</sup> and (2.6 <sup>±</sup> 0.2) mgN L−1, respectively), indicating a limitation of this nutrient in the last days of assay 5. As for nitrogen concentration, phosphorus concentration also decreased within the cultivation time (Figure 3B), but in a lesser extent, which is related with microalgal nutritional requirements, as given by its typical elemental biochemical composition: CO0.48H1.83N0.11P0.01 [24]. The reduction observed in nitrogen and phosphorus concentration in the studied effluent (raw or diluted) shows that *C. vulgaris* can promote an efficient uptake of both nutrients. However, except for nitrogen concentration in assay 5, total depletion of these nutrients did not occur after the 11 days of culturing, reiterating what was stated in relation to cell growth, that the cultures could be extended for an increased period to further improve nutrients removal efficiencies. Another similarity with the microalgal growth parameters already described is the higher variations in nitrogen and phosphorus concentrations observed in the experiments where the effluent was previously diluted (assays 2–5), which indicate that these conditions were more favorable for *C. vulgaris* photosynthetic activity.

Nitrogen and phosphorus removal parameters are presented in Figures 4 and 5, respectively. As with microalgal growth parameters, a general increase in nutrients removal efficiencies was observed from assay 1 to 5, with values ranging from (24 ± 10)% to (80 ± 4)% for nitrogen (Figure 4A) and from (13.0 ± 0.9)% to (54 ± 1)% for phosphorus (Figure 5A). However, Figure 4A shows that there was no statistical difference (*p* > 0.05) in nitrogen removal efficiency between assays 4 and 5, which can be explained by the low concentration achieved in the assay 5 (the one corresponding to the most diluted effluent) that might have been limiting for microalgal growth. In fact, according to Table 2, maximum biomass concentration achieved in assay 4 was higher than that in assay 5, indicating that the highest dilution applied in this study may have contributed to nitrogen limitation to *C. vulgaris*, with effects on their growth and nutrients removal parameters. Regarding nitrogen removal rates (Figure 4B) and mass removal (Figure 4C), the highest values were determined in assays 3 and 4 and no statistical differences were observed (*<sup>p</sup>* <sup>&</sup>gt; 0.05): (i) average removal rates were (1.31 <sup>±</sup> 0.07) mgN <sup>L</sup>−<sup>1</sup> <sup>d</sup>−<sup>1</sup> and (1.26 <sup>±</sup> 0.08) mgN <sup>L</sup>−<sup>1</sup> <sup>d</sup><sup>−</sup>1, respectively; and (ii) mass removal values were (14.4 <sup>±</sup> 0.8) mgN <sup>L</sup>−<sup>1</sup> and (13.9 <sup>±</sup> 0.9) mgN L−1, respectively. These results are in accordance with maximum biomass concentration achieved and indicate higher photosynthetic activity of *C. vulgaris* in these intermediate conditions. A different behavior was observed for phosphorus. In this case, average removal rates (Figure 5B) and mass removal values (Figure 5C) determined for assays 1 to 4 were not statistically different (*p* > 0.05), but values determined for assay 5 were statistically higher (*p* < 0.05), reaching an average removal rate of (0.20 <sup>±</sup> 0.01) mgP <sup>L</sup>−<sup>1</sup> <sup>d</sup>−<sup>1</sup> and a mass removal of (2.2 <sup>±</sup> 0.1) mgP <sup>L</sup><sup>−</sup>1.

**Figure 3.** Temporal variation of (**A**) nitrogen (nitrate + nitrite) and (**B**) phosphorus (phosphate) concentration determined in *C. vulgaris* cultures grown in raw and diluted secondary-treated paper industry effluent (Assays: 1 ,, 2 ,, 3 ,, 4 and 5 ). Error bars correspond to the standard deviation of the mean obtained from two independent experiments.

**Figure 4.** *Cont.*

**Figure 4.** Nitrogen (nitrate + nitrite) removal parameters obtained by *C. vulgaris* cultures grown in raw and diluted secondary-treated paper industry effluent (assays 1–5): (**A**) removal efficiency (%RN); (**B**) average removal rate (RRN); and (**C**) mass removal (RN). Error bars correspond to the standard deviation of the mean obtained from two independent experiments. Mean values sharing at least one common letter (shown above the bars) are not statistically different (*p* > 0.05).

**Figure 5.** Phosphorus (phosphate) removal parameters obtained by *C. vulgaris* cultures grown in raw and diluted secondary-treated paper industry effluent (assays 1–5): (**A**) removal efficiency (%RP); (**B**) average removal rate (RRP); and (**C**) mass removal (RP). Error bars correspond to the standard deviation of the mean obtained from two independent experiments. Mean values sharing at least one common letter (shown above the bars) are not statistically different (*p* > 0.05). <sup>85</sup>

Nutrients removal from paper industry effluents has already been reported in the literature. Table 3 highlights nitrogen and phosphorus removal efficiencies and removal rates obtained in these studies. According to these data, removal efficiencies reported by Tao, et al. [19] and Gentili [2] are significantly higher than those obtained in this study, whereas values reported by Usha, et al. [1] were closer to those obtained in the present study, especially in assays 3–5. The lower removal efficiencies obtained in this study when compared with those reported by Tao, et al. [19], may be associated with the higher N:P molar ratio used in the reference study, which was ~ 66:1. On the other hand, the higher removal efficiencies reported by Gentili [13] may be associated with the use of other effluents to achieve the dual role of providing the required nutrients for microalgal growth while contributing to a reduction in the toxicity of the paper industry effluent. Another explanation for the increased efficiencies obtained in these studies is the nitrogen source used. As in the present study, Usha, et al. [1] cultivated microalgae in an effluent with nitrate–nitrogen as the main nitrogen source. On the other hand, Tao, et al. [19] tested an effluent with ammonium as the main nitrogen source (digestate obtained from the treatment of a pulp and paper industry effluent) and Gentili [13] evaluated this treatment with both nitrogen forms present. According to several studies, although nitrate–nitrogen is the most thermodynamically stable form (and the most commonly found in aquatic environments), ammonia is directly assimilated and converted into proteins by microalgae, while nitrate must be reduced to nitrite and then to ammonia before being assimilated by microalgal cells [25]. However, for an adequate comparison of nutrients removal performance, it is important to determine the average removal rate, as this parameter takes into account initial nutrients concentrations and cultivation/treatment time. Comparing average removal rates obtained in the present study and in the reference studies, values in the same order of magnitude were obtained, except in what concerns ammonium–nitrogen removal in the studies performed by Tao, et al. [19] and Gentili [2]. In these cases, the higher removal rates obtained may be associated with the higher ability of microalgae to assimilate ammonium–nitrogen than nitrate–nitrogen. Considering values of average RR, it is possible to conclude that promising results were obtained in this study. Moreover, differences found in experimental conditions used in this study and in the studies reported in the literature demonstrate that these results can be significantly enhanced. Besides increasing N:P molar ratio and providing an ammonium–nitrogen source, the increase of light supply should also be considered, as values reported in the literature correspond to cultures grown under light intensities of 130–800 μmol m−<sup>2</sup> s<sup>−</sup>1, whereas results reported in the present study were obtained with light intensities of 30–40 μmol m−<sup>2</sup> s<sup>−</sup>1.

In summary, the results obtained in this study for both nitrogen and phosphorus removal evidence that the remediation of paper industry effluents using microalgae is possible, provided that it is properly diluted to avoid inhibitory effects related to the presence of strong color or high concentrations of toxic compounds, typically associated with effluents resulting from this industrial sector [19,20]. Considering the results obtained for nitrogen removal, the dilution of the effluent to the concentrations present in assays 3 and 4 is the most adequate. In these conditions, nitrogen concentrations were significantly reduced, reaching (7.1 <sup>±</sup> 0.7) mgN <sup>L</sup>−<sup>1</sup> and (2.81 <sup>±</sup> 0.05) mgN <sup>L</sup><sup>−</sup>1, respectively (which corresponds to the highest average removal rates: (1.31 <sup>±</sup> 0.07) mgN <sup>L</sup>−<sup>1</sup> <sup>d</sup>−<sup>1</sup> and (1.26 <sup>±</sup> 0.09) mgN <sup>L</sup>−<sup>1</sup> <sup>d</sup><sup>−</sup>1, respectively). Regarding phosphorus removal, the highest removal rate was obtained for the conditions tested in assay 5: (0.20 <sup>±</sup> 0.01) mgP <sup>L</sup>−<sup>1</sup> <sup>d</sup><sup>−</sup>1.


**Table 3.** Comparison between nutrients removal efficiencies (%R, in %) and average removal rates (RR, in mg L−<sup>1</sup> d<sup>−</sup>1) obtained in this study and other studies reporting microalgal growth in effluents resulting from pulp and paper industries.

Despite the promising nitrogen and phosphorus removal rates, the results obtained in this study demonstrated that the cultures were limited by nitrogen, as nitrogen and phosphorus were assimilated by *C. vulgaris* at a N:P molar ratio ranging from 10:1 to 24:1. Considering these results and the N:P molar ratios used in this study (between 6:1 and 9:1), nutrients uptake could be enhanced by increasing nitrogen supply. Another alternative to achieve an adequate N:P molar ratio and reduce the toxicity of this effluent would be to dilute it with other effluents, as proposed in other studies [13]. Finally, the remediation process could be further improved by modulating microalgal cultivation conditions. According to Gonçalves, et al. [25], light conditions, temperature, and pH are also important parameters that can influence microalgal growth and, hence, the efficiency of the bioremediation process. Thus, from the prospecting of this work, other studies evaluating these parameters should be carried out to further improve nitrogen and phosphorus uptake from paper industry effluents.

#### **4. Conclusions**

This study showed the feasibility of using *C. vulgaris* for the bioremediation of a paper industry effluent fortified with a nitrogen source, targeting phosphorus removal. *C. vulgaris* was able to grow in all studied effluent conditions (in non-diluted and diluted ones). However, it was possible to conclude that growing on non-diluted effluent resulted in lower biomass productivities, which was also reflected in nitrogen and phosphorus removal efficiencies. From microalgal growth and nitrogen removal points of view, the effluent dilutions used in assays 3 and 4 (intermediate dilutions) seem to be the most adequate, as microalgal growth was not inhibited in these conditions and nitrogen mass removal was quite satisfactory, achieving final concentrations of (7.1 <sup>±</sup> 0.7) mgN L−<sup>1</sup> and (2.81 <sup>±</sup> 0.05) mgN L−1, respectively. Regarding phosphorus removal, concentrations achieved in the last day of culturing in assays 3 and 4 were higher ((4.63 <sup>±</sup> 0.04) mgP <sup>L</sup>−<sup>1</sup> and (2.940 <sup>±</sup> 0.005) mgP <sup>L</sup><sup>−</sup>1, respectively) than the one obtained in assay 5 ((1.85 <sup>±</sup> 0.02) mgP <sup>L</sup><sup>−</sup>1). However, the results obtained in assay 5 suggest a growth limitation, mainly related to nitrogen concentration. Accordingly, the obtained results indicate that these

values can be further improved by studying different N:P molar ratios, different microalgal cultivation conditions, dilution with other effluents, among others. Improving the remediation performance can significantly contribute to the development of an effective microalgae–based remediation process of pulp and paper industry effluents.

**Author Contributions:** Conceptualization, B.P., A.L.G., V.J.PV. and J.C.M.P.; methodology, B.P., A.L.G. and J.C.M.P.; investigation, B.P. and A.F.E.; resources, V.J.P.V. and J.C.M.P.; data curation, B.P., A.L.G., A.F.E., V.J.P.V. and J.C.M.P.; writing (original draft preparation), B.P.; writing (review and editing), B.P., A.L.G., A.F.E., S.M.A.G.U.d.S., A.A.U.d.S., V.J.P.V. and J.C.M.P.; supervision, A.L.G., S.M.A.G.U.d.S., A.A.U.d.S., V.J.P.V. and J.C.M.P.; project administration, S.M.A.G.U.d.S., A.A.U.d.S., V.J.P.V. and J.C.M.P.; funding acquisition, S.M.A.G.U.d.S., A.A.U.d.S., V.J.P.V. and J.C.M.P. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by: (i) Base Funding—UIDB/00511/2020 of the Laboratory for Process Engineering, Environment, Biotechnology, and Energy—LEPABE—funded by national funds through the FCT/MCTES (PIDDAC); (ii) Base Funding—UIDB/50020/2020 of the Associate Laboratory LSRE-LCM—funded by national funds through FCT/MCTES (PIDDAC); (iii) Project PTDC/BTA-BTA/31736/2017—POCI-01-0145-FEDER-031736—funded by FEDER funds through COMPETE2020—Programa Operacional Competitividade e Internacionalização (POCI) and with financial support of FCT/MCTES through national funds (PIDDAC); and (iv) the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior—Brasil (CAPES)—Finance Code 001. V.J.P. Vilar acknowledges the FCT Individual Call to Scientific Employment Stimulus 2017 (CEECIND/01317/2017). J.C.M. Pires acknowledges the FCT Investigator 2015 Programme (IF/01341/2015).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 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 (http://creativecommons.org/licenses/by/4.0/).

### *Review* **Arsenic Contamination of Groundwater and Its Implications for Drinking Water Quality and Human Health in Under-Developed Countries and Remote Communities—A Review**

**Samuel B. Adeloju 1,2,\*, Shahnoor Khan 1,3 and Antonio F. Patti <sup>1</sup>**


**Featured Application: Groundwater contamination is a major global issue. A good understanding of the associated chemistry and fate of a contaminant such as arsenic in groundwater is important for minimizing or avoiding potential health, social and economic implications when used as a water source for human consumption.**

**Abstract:** Arsenic is present naturally in many geological formations around the world and has been found to be a major source of contamination of groundwater in some countries. This form of contamination represents a serious threat to health, economic and social well-being, particularly in under-developed countries and remote communities. The chemistry of arsenic and the factors that influence the form(s) in which it may be present and its fate when introduced into the environment is discussed briefly in this review. A global overview of arsenic contamination of groundwater around the world is then discussed. As a case study, the identified and established causes of groundwater contamination by arsenic in Bangladesh is highlighted and a perspective is provided on the consequential health, agricultural, social and economic impacts. In addition, the relevant removal strategies that have been developed and can generally be used to remediate arsenic contamination are discussed. Also, the possible influence of groundwater inorganic compositions, particularly iron and phosphate, on the effectiveness of arsenic removal is discussed. Furthermore, some specific examples of the filter systems developed successfully for domestic arsenic removal from groundwater to provide required potable water for human consumption are discussed. Lastly, important considerations for further improving the performance and effectiveness of these filter systems for domestic use are outlined.

**Keywords:** arsenic; groundwater; contamination; water quality; domestic filter systems; health effects; treatment methods; Bangladesh

#### **1. Introduction**

Arsenic (As) is a chemical element which occurs naturally and is commonly present in the earth's crust [1]. It has been found in air, biota, water, soil and rocks [1,2]. In particular, arsenic present in an aqueous medium is of most concern because of its likely detrimental impact on plants, animals and humans. When present in high concentrations in drinking water, arsenic has been found to adversely affect human health [3] and this topic has been the subject of several recent reviews [4–6]. In this comprehensive review, we have additionally provided a detailed evaluation of treatment options for arsenic removal from water, with due consideration for affordable small scale technologies that can be easily implemented in developing countries such as Bangladesh, India, Nepal and Pakistan.

**Citation:** Adeloju, S.B.; Khan, S.; Patti, A.F. Arsenic Contamination of Groundwater and Its Implications for Drinking Water Quality and Human Health in Under-Developed Countries and Remote Communities—A Review. *Appl. Sci.* **2021**, *11*, 1926. https://doi.org/ 10.3390/app11041926

Academic Editors: Bart Van der Bruggen and Dino Musmarra

Received: 7 January 2021 Accepted: 16 February 2021 Published: 22 February 2021

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

**Copyright:** © 2021 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/).

Arsenic has been introduced to the environment through various natural and anthropogenic sources [7]. The primary natural source of arsenic is from As-enriched minerals [2]. Typical examples of natural sources include volcanoes and eroded arsenic bearing rocks, such as arsenopyrite (FeAsS), lollingite (FeAs2), orpiment (As2S3) and realgar (AsS) [8]. Under oxidizing and reducing conditions, arsenic can be mobilized at pH 6.5–8.5 [3,9] which is a common pH range in groundwater [9]. On the other hand, the common anthropogenic sources include agriculture, livestock and industrial manufacturing [10]. It is also widely used for the manufacture of glassware, industrial chemicals, copper, lead alloys and pharmaceuticals [10]. Industrial processes, such as smelting of iron ores, mining, pulp and paper production, cement manufacture, burning of fuels and wastes are known sources for the release of arsenic into the environment [8]. However, the focus of this review is on the contamination of groundwater by arsenic from predominantly natural sources.

Arsenic is a group Va element on the periodic table and is classified as a metalloid. In the environment, it exists in several oxidation states as As(3-) (arsine), As(0) (arsenic), As(3+) (arsenite) and As(5+) (arsenate) [11]. In natural waters, the common soluble arsenic species are the inorganic oxyanions of As(III) or As(V) [12]. As(III) species include As(OH)3, H2AsO3 −, HAsO3 <sup>2</sup><sup>−</sup> and AsO3 <sup>3</sup><sup>−</sup> [13,14]. As(V) species are AsO4 <sup>3</sup>−, HAsO4 <sup>2</sup>−, H2AsO4 – and H3AsO4 [13,14]. Arsenic (III) forms complexes preferentially with oxides and nitrogen [14]. On the other hand, As(V) forms complexes with sulfides [14].

There are also organic arsenic species such as arseno-sugars dimethyl arsenic acid (DMAA) and monomethyl arsenic acid (MMAA) [15]. However, in drinking water treatment, these species are not of significant concern due to the limited effect they have on human health, and their ease of elimination from the body [15].

Under reducing anaerobic conditions, the dominant species are trivalent arsenic [9], while in oxygen rich aerobic conditions pentavalent species are dominant [9,11,12]. However, the key considerations for controlling the arsenic species present under various conditions are redox potential (Eh) and pH [9,16–18]. Figure 1 shows that, under oxidizing conditions, H2AsO4 − is the dominant arsenate species at low pH (pH < 6.9), whereas HAsO4 <sup>2</sup><sup>−</sup> is the dominant arsenate species at a higher pH, as illustrated in Figure 1c [9,16,17]. On the other hand, H3AsO3 <sup>0</sup> is the dominant arsenite species at pH < 9.2 when operating in reducing conditions, as illustrated in Figure 1b.

**Figure 1.** pH dependence of the aqueous speciation of As(III) and As(V) species. (**a**) redox potential (Eh)–pH diagram [9,17], (**b**) arsenite species and (**c**) arsenate species [16]. Operating conditions: 25 ◦C and 1 bar total pressure. Reproduced with permission from Elsevier and RSC.

The presence of nitrate in groundwater has been proposed as a likely significant contributing mechanism for the oxidation of arsenic to As(V) [18]. Nitrate can act as a terminal electron acceptor and this mechanism is likely to be more prevalent under anoxic conditions.

Reported cases of groundwater contamination by arsenic are more diverse than commonly realized and extend from under-developed to developed countries, including Argentina, Bangladesh, Canada, Chile, China, Hungary, India, Japan, Mexico, Nepal, Poland, Taiwan and the USA [15]. Figure 2 shows where groundwater or surface water contamination by arsenic has been reported. Croatia, Hungary and Serbia are three of the countries in Europe where very high concentrations of arsenic in groundwater have been reported. In Hungary, the arsenic concentrations in groundwater were found to exceed the WHO (10 μg/L) and EC guidelines (7.5 μg/L, EC Directive 2006/118/EC) by several times [19]. The concentrations of arsenic in various groundwater samples collected in Hungary and Romania ranged from <0.5 to 240 μg/L [20]. More distinctly, the highest arsenic concentrations (23 to 208 μg/L, mean 123 μg/L) were obtained in waters dominated by methanogenesis, whereas waters dominated by sulfate reduction gave lower arsenic concentrations (<0.5 to 58 μg/L, mean 11.5 μg/L) [20]. In Serbia, the extent of arsenic contamination is not fully resolved and yet to be determined [15]. As illustrated in Figure 2, the most affected countries in the Americas are Argentina, Chile, Mexico and the United States. In Latin America, the estimated number of people affected by arsenic contamination >50 μg/L is at least four million [21]. Some wells in Argentina, Bolivia and Peru were found to contain extremely high arsenic concentrations at high mg/L levels [21].

**Figure 2.** Countries affected by arsenic contamination of groundwater or surface waters [15]. Reproduced with permission from IJAET.

The global variation of arsenic concentrations in groundwater is summarized in Table 1. Among the affected countries, Bangladesh and West Bengal in India have the largest population at risk of exposure to arsenic contamination [7]. This is why we chose Bangladesh as a case study to highlight the extent of groundwater contamination by arsenic. The people most affected in Bangladesh and West Bengal reside where there is no access to integrated urban water supply systems, mostly in rural areas. Consequently, millions of people living in rural areas in these countries are regularly drinking water contaminated with arsenic. A relatively large number of people in this region have already been identified as having arsenic related disease symptoms [15].

UNICEF reported that 1.4 million tube wells out of the 4.75 million (around 30%) in Bangladesh contained arsenic concentration above 50 μg/L [22]. A household drinking water quality survey conducted by UNICEF in Bangladesh found that 12.6% of drinking water samples exceeded the country's arsenic standard for drinking water [22]. This represents approximately 22 million Bangladesh people at risk of adverse arsenic exposure [22]. Furthermore, in West Bengal (India), an estimated 6 million people were found to have been exposed to relatively high arsenic concentrations, ranging from 50 to 3,200 μg/L [23,24].


**Table 1.** Global variation of arsenic concentrations in groundwater.

Adapted and modified from [3], \* [15] and # [25].

Usually after a few years of extraction of groundwater, a considerable increase in the concentration of arsenic in wells has been observed [15]. Future increases in arsenic contamination issues in drinking water are expected despite the establishment of a stringent standard for arsenic in drinking water and this may spread to other countries [15]. Due to the greater risk that arsenic contamination of groundwater poses to the Bangladeshi people, it is useful to take a closer look at the widespread nature of the problem in this country.

#### **2. Case Study—Arsenic Contamination of Groundwater in Bangladesh**

#### *2.1. Background*

Bangladesh is a tropical country with a total area of about 147,570 km2 and an estimated population of 160 million as of 2016 [26]. About 58% of the surface area is arable land and about 11% comprises forests and woodlands. The contribution of the agricultural sector to national GDP (Gross Domestic Product) is about 18%, while about 72% of its people are based in rural settings [26]. The per capita income in 2015–2016 is US\$ 1466 [26].

Although there is abundant groundwater in Bangladesh and the productivity of the aquifers are very high, variable water tables are observed in different parts of the country, more commonly shallow and below the ground surface by 1–10 m [27]. Groundwater is generally free from pathogenic microorganisms and, if not contaminated, can be reasonably safe to use. For these reasons groundwater has remained an appealing and readily available reserve for drinking water in many countries [27]. This has resulted in the widespread use of groundwater as the main source of drinking water, particularly as tube wells for the past forty years or more. The tube wells have contributed significantly to the reduction of mortality rate of diarrheal diseases [28]. Bangladesh achieved a remarkable success by providing 97% of the rural population with tube well water [28].

However, the coverage for safe drinking water has been significantly reduced from 97% to 74% as a consequence of the extensive contamination of groundwater with arsenic [28]. Consequently, the mortality rate in Bangladesh was increased as a result of arsenic contamination of the drinking water [29]. About 12.6% of its population are still accessing this contaminated water [22]. This has led to the recognition of arsenic contamination as a catastrophic issue in Bangladesh [22]. To understand the cause of this problem, a knowledge of the geological characteristics of Bangladesh is useful.

#### *2.2. Geology of Bangladesh*

Young (Holocene) alluvial and deltaic sediments dominate the geology of Bangladesh. The major river systems of the Bengal Basin are responsible for the deposition of these sediments [27]. Most of these sediments, of several hundreds of meters in thickness, have been deposited within the last 6000–10,000 years [27]. The Basin is now recognized in the world among the most rapidly developed delta systems [27]. In the north, the surface sediments consist mainly of coarse-grained mountain-front alluvial fan deposits, while alluvial sands and silts represent the main composition of sediments from central Bangladesh. The predominant composition of the sediments in the south are deltaic silts and clays. On approaching the southern part of Bangladesh, the deposits tended to be increasingly fine-grained [27]. Figure 3 illustrates the surface geological units of Bangladesh. Evidently, the surface geological characteristics are very diverse in nature.

#### *2.3. Mobilization of Arsenic in Groundwater in Bangladesh*

The two most widely discussed hypotheses for the sources of arsenic contamination of groundwater in Bangladesh are the pyrite oxidation hypothesis and the iron oxyhydroxide reduction hypothesis [6,30].

The "Barind and Modhupur Tracts" made up of up-faulted terraces of older (Pleistocene) sediments are the key characteristics of the surface geology in north-central Bangladesh. Unlike the surrounding alluvium deposits, these sediments have experienced significantly more weathering [27]. The younger alluvial sediments in these tracts are present at depths ranging or greater than 50–200 m. However, for the older sediments, the extent of its distribution is still relatively unknown [27]. In south-east Bangladesh, older (Tertiary) sediments dominate the geology of the Chittagong Hill Tracts and their main composition are sandstone, silt and limestone [27]. Within 20–80 m depth, the

sediment was found to be very rich in arsenopyrite [31]. The granitic and metamorphic processes of the Himalayas was the original source of the sediment [31]. After the introduction of the sediment into the Ganges delta, it became part of the aquifers [31]. Additional arsenic is retained as an adsorbed coating with ferric oxy-hydroxide on the sediments [32]. It is understood that the transport and deposition of arsenic and ferric oxy-hydroxide in the Ganges delta along with abundant organic matter resulted from the oxidation of arsenopyrite [33,34]. Figure 4 illustrates the arsenic contaminated areas in Bangladesh [24]. Obviously, the arsenic concentrations present in many areas are considerably above 50 μg/L.

**Figure 3.** Surface geological units of Bangladesh [27]. Reproduced with permission from BGS.

**Figure 4.** Arsenic contaminated areas in Bangladesh [24]. Reproduced with permission from BGS.

2.3.1. Pyrite Oxidation Hypothesis

Studies have shown that there is a high level of arsenopyrite (FeAsS) in the alluvial regions of Bangladesh [31,35]. The basis of the pyrite oxidation hypothesis is that the oxidation of arsenopyrite resulted in the release of arsenic into groundwater [36–39]. This process is understood to be aided by the invasion of the aquifers by atmospheric oxygen when the water table is lower than these deposits, resulting in its diffusion into

the pore space and groundwater. Consequently, the reaction between the arsenopyrite and available oxygen resulted in a water-soluble form of arsenic which was consequently released into the groundwater [30].

The associated reactions for this process are:

$$4\text{FeAsS} + 11\text{O}\_2 + 6\text{H}\_2\text{O} \rightarrow 4\text{FeSO}\_4 + 4\text{H}\_2\text{AsO}\_3^- + 4\text{H}^+$$

$$4\text{FeAsS} + 13\text{O}\_2 + 6\text{H}\_2\text{O} \rightarrow 4\text{FeSO}\_4 + 4\text{H}\_2\text{AsO}\_4^- + 4\text{H}^+$$

The above hypothesis is supported by the absence of arsenic-affected people before irrigation became more intensive at the beginning of the 1980s. Consequently, the total irrigated area which used groundwater increased significantly from 41% to 71% within 1982/1983 and 1996/1997, respectively, while in the same period the use of surface water decreased considerably from 59% to 29% [30].

Based on the pyrite oxidation hypothesis, the lowering of the water table resulted from the considerable use of groundwater, and, in turn, led to groundwater contamination with arsenic. The arsenic mobilization proposed by this hypothesis is widely accepted as the main mechanism for intrusion of arsenic into groundwater in Bangladesh [36,40,41]. However, the hypothesis is not yet fully validated as there is a need for further hydrological and geochemical data [42].

#### 2.3.2. Iron Oxyhydroxide Reduction Hypothesis

Another suggestion proposed as the source of groundwater contamination with arsenic in Bangladesh is based on the so-called iron oxyhydroxide reduction hypothesis. This hypothesis is based on the premise that the source of arsenic was from the Ganges region, upstream of Bangladesh, where arsenic sulfide minerals followed by abrasion were carried by water about 1.6–1.8 million to 10,000 years ago during the late Pleistocene age [43]. As these arsenic-containing minerals travelled down the Ganges, arsenic was adsorbed to iron oxyhydroxide (FeOOH). The arsenic-rich iron oxyhydroxides were deposited at the Gangetic delta, and formed an alluvial aquifer. Various processes, including burial of vegetation, agriculture and floods, led to the introduction of organic carbon into the aquifer [8]. This organic carbon provided food for bacteria and the presence of methane in the water indicates that organic matter is being utilized in the aquifer by anaerobic bacteria [30]. Consequently, the redox potential of the groundwater is lowered by this biological process. Due to this reducing environment, the iron oxyhydroxide is broken down, resulting in the introduction of adsorbed arsenic into the groundwater [36,39,43–45]. The adsorption of arsenic by hydrous iron oxyhydroxide is supported by the following reactions [8]:

$$\text{Fe (OH)}\_{3} + \text{H}\_{3}\text{AsO}\_{4} \rightarrow \text{FeAsO}\_{4} \cdot 2\text{H}\_{2}\text{O} + \text{H}\_{2}\text{O}$$

$$\equiv \text{FeOH}^{\text{o}} + \text{AsO}\_{4}^{3-} + 3\text{H}^{+} \rightarrow \equiv \text{FeH}\_{2}\text{AsO}\_{4} + \text{H}\_{2}\text{O}$$

$$\equiv \text{FeOH}^{\text{o}} + \text{AsO}\_{4}^{3-} + 2\text{H}^{+} \rightarrow \equiv \text{FeHAsO}\_{4}^{-} + \text{H}\_{2}\text{O}$$

The reduction of hydrous iron oxides can occur by the following reaction:

$$\text{Fe(OH)}\_{3} + \text{e}^{-} \rightarrow \text{Fe}^{2+} + \text{3OH}^{-}$$

Therefore, the presence of anoxic conditions is necessary for arsenic-rich iron oxyhydroxide to be reduced and, in turn, release arsenic. This process mobilizes iron and its adsorbed load into the groundwater. Some researchers have accepted the arsenic mobilization based on the iron oxyhydroxide reduction hypothesis as the main mechanism for groundwater contamination by arsenic in Bangladesh [27,36,42,46]. However, a more comprehensive sampling and systematic analysis of ferric oxyhydroxide in the areas affected is needed to validate the reduction hypothesis [27,42].

Furthermore, it may also be possible for both of the mechanisms discussed above to contribute to arsenic release. This is in view of the fact that the pyrite oxidation occurs under an oxic condition, while iron oxyhydroxide reduction occurs under an anoxic condition.

#### **3. Effects of Arsenic Contamination**

Groundwater contamination by arsenic in Bangladesh has had adverse effects on human health, agriculture, social well-being and the economy of the country. To highlight the seriousness of these impacts, each of these is discussed in more details below.

#### *3.1. Effects on Health*

Groundwater contamination by arsenic has resulted in numerous serious health consequences in Bangladesh [47–49]. Globally, arsenic contamination in drinking water has reportedly led to the exposure of about 150 million people leading to serious health effects [50]. Of these, about 110 million people live in Bangladesh, Cambodia, China, India, Laos, Myanmar, Nepal, Pakistan, Taiwan and Vietnam [51]. In addition, many people are affected through consumption of arsenic-contaminated foods produced from the use of arsenic contaminated groundwater [50,51]. It has been found that rice grains contribute about half of the daily intake of arsenic in Bangladesh [52]. A survey conducted within arsenic-affected villages in Bangladesh collected hair, nail, urine and skin-scale samples totaling more than 10,000 which were subsequently analyzed for arsenic [53]. The results of the survey indicated that the arsenic concentrations in 93.8% and 95.1% of nails and urine samples, respectively, exceeded the normal level. For nail, the acceptable arsenic concentration range is 0.43–1.08 mg/kg [54], while for urine the range is 0.005–0.04 mg/day [55]. Furthermore, 83.2% and 97.4% of hair and skin-scale samples, respectively, gave arsenic contents that far exceeded the toxic level of 1 mg/kg [56]. These results therefore revealed a serious impact on health that has arisen from human exposure to arsenic through groundwater contamination in Bangladesh.

A recent comprehensive overview on the adverse effects from inorganic arsenic exposure details the major health issues [57]. Carcinogenic effects are particularly prevalent and the most common consequences resulting from poisoning from drinking arsenic contaminated water are skin diseases, including hyperkeratosis, keratosis, leuco-melanosis and melanosis (hyper pigmentation) [57–64]. Arsenicosis is initially manifested by melanosis which usually occurs all over the body [57,62]. The hardening of the melanosis spots leads to the commencement of keratosis [62]. A long-term exposure to arsenic results in the hyperkeratosis of the palms and soles [59]. Other associated health effects resulting from ingestion of arsenic include cardiac failure, chromosomal abnormality, cirrhosis, diabetes mellitus, gangrene, goiter, hypertension (high blood pressure), liver enlargement, myocardial degeneration, peripheral neuropathy and skin cancers [59,62,63].

Also, the development of cognitive and psychological functions in children has been shown to be affected by extended consumption of arsenic contaminated drinking water [65]. Furthermore, areas where highly contaminated groundwater is used have been shown to be more prone to higher fetal loss and infant deaths [66].

#### *3.2. Effects on Agriculture*

Groundwater is the main source of water used for agriculture in Bangladesh. Consequently, when contaminated with arsenic, it can result in the introduction of arsenic into agricultural soils and crops, such as rice and vegetables [51,67,68]. For the same reason, it has also been found in many areas in Bangladesh that high arsenic concentrations are present in the agricultural soils and the usage of groundwater contaminated with arsenic for irrigation has been identified as the main cause [50,52].

A main consequence is the dual arsenic accumulation in vegetables and rice grains from both soils and irrigation water [30,50,51,69–71]. More seriously, the phytotoxic effects of arsenic can result in a significant reduction of crop yields [52]. The variation of arsenic contents of selected foods from different countries is demonstrated by the data

in Table 2. In general, arsenic concentrations in these foods are lowest when obtained in countries with little or no exposure to arsenic contamination of water source.

Rice, which is a staple part of the diet in Bangladesh, has been extensively studied in the context of mitigating and reducing the uptake of arsenic during its cultivation. An excellent review on this topic has been recently published [72]. The authors have indicated that by employing water management, physico-chemical and biological strategies, either alone or combined, the uptake of inorganic and methylated arsenic species in rice cultivation can be successfully decreased. These approaches may also be adapted for other crops. The adoption of these strategies is a major challenge for countries like Bangladesh, where socio-economic factors can be a hindrance.


**Table 2.** Concentrations of arsenic in selected foods from different countries.

<sup>a</sup> Samples collected from arsenic-affected area and <sup>b</sup> marine species. Adapted and modified from [73].

#### *3.3. Social Effects*

The prevalence of arsenic poisoning has triggered many social implications [48,49,74,75]. Relatives, friends and neighbors often ostracize the people affected by arsenic poisoning. There is a general tendency to avoid or discourage the arsenic-affected people from being seen in public. More concerning, school attendance by their children is usually prohibited, while attendance at work places and public meetings is not encouraged for the adults [76].

For women, arsenic poisoning has more dire social consequences [76]. Young women who have had arsenicosis are often constrained from getting married. Even married women who have arsenic related diseases are also socially ostracized. Evidently, the males, females and children who are affected by arsenic poisoning are severely disadvantaged socially [76].

#### *3.4. Economic Impact*

There is a direct relationship between the economic impact of suffering from arsenicosis and the social impact experienced from the disease [48,49,74,75,77]. Usually once a family member becomes sick from arsenic poisoning, various coping mechanisms are experienced. This may involve selling of assets, reduction of basic needs, reduction of access to education and the associated burden of financial loans [76]. Consequently, a large-scale poisoning of the population can impact the nation's economy, requiring the Government, in extreme cases, to reduce its social and economic development programs as a basis for dealing with the disaster.

Attempts to minimize or eliminate the impacts of groundwater contamination with arsenic have resulted in various strategies for arsenic removal from groundwater with a varying level of successes. Nevertheless, there is still considerable research being undertaken in this area to achieve efficient and reliable removal of arsenic from groundwater to ensure safe human consumption. Some of these removal processes and strategies are discussed below.

#### **4. Arsenic Removal Processes**

A major approach for reducing arsenic poisoning is by treatment of arsenic-contaminated water and the new and emerging treatment technologies have received considerable evaluation and have also been recently reviewed [47,78–80]. The major categories of treatment methods that have been used for this purpose are described below [81]:


The following sub-sections provide more detailed discussion of some of these treatment methods.

#### *4.1. Biological Process*

Numerous biological treatment options for removal of metals from drinking water are available [99] and the biological treatment of groundwater for arsenic removal employs naturally occurring microorganisms, such as *Gallionellaferruginea* and *Leptothrixochracea* [100]. The addition of these organisms to the groundwater results in the formation of new iron oxide precipitates on the filter, which subsequently removes arsenic from the water by adsorption. Under optimized conditions, As(III) is oxidized by these microorganisms, enabling removal of up to 95% of arsenic from waters which contain 200 μg/L of arsenic [100]. This process also enabled the removal of As(V). For a comprehensive overview of biological transformations of arsenic, which includes possible detoxification mechanisms, refer to a recent review on this topic [101].

#### *4.2. Precipitative Processes*

This category of arsenic removal processes includes coagulation and filtration [87], coagulation-assisted microfiltration [102], enhanced coagulation [103] and lime softening [104]. These are very popular processes that require the addition of either aluminum sulfate, ferric chloride or ferric sulfate as a coagulant to change either the chemical or physical properties of dissolved colloidal or suspended matter [105]. To promote rapid settling out of the particles by gravity, the enhancement of agglomeration is used in some cases, otherwise filtration is used to remove the particles [105]. Agglomeration is usually accomplished by changing the surface charge properties of solids with coagulants to enable formation of a flocculated precipitate. The resulting products of this process are larger particles or flocs that filter or settle out more easily under gravity [106].

Greater than 90% As(V) removal has been successfully achieved with coagulation processes [107]. The use of coagulation with alum, ferric chloride and ferric sulfate for removal of As(III) is far less efficient than the removal of As(V) [107]. To ensure efficient arsenic removal, As(III) is usually oxidized to As(V) prior to coagulation. Effective As(V) removal is usually accomplished at pH 7.6 or less with iron and aluminum coagulants. However, in terms of stability, iron coagulants offer more advantage than aluminum coagulants when operating within a pH range of 5.5 to 8.5 [9]. The key factors that influence the choice of the optimum coagulant dose are the water quality and the arsenic concentration in the treated water. The overall cost of the water treatment process can be significantly increased if further adjustment of the pH is necessary to achieve a more effective removal of arsenic [106].

Some field studies have demonstrated the effectiveness of coagulation/filtration in reducing the arsenic level to below 5 μg/L [106]. However, under optimum operating conditions, the achievement of a residual arsenic level of less than 3 μg/L is possible [107].

#### *4.3. Membrane Processes*

Siddique et al. [108] have recently reviewed the application of nanofiltration membrane technologies and their advantages and disadvantages relative to other methods. The membrane processes used for arsenic removal mainly include reverse osmosis [94,95] and electrodialysis [96]. While these processes are effective, they are generally more costly. Nonetheless, they are capable of removing arsenic through filtration, electric repulsion, and adsorption of arsenic-bearing compounds. The size exclusion property of a chosen membrane enables rejection of arsenic compounds larger than its pore size [107]. Besides the size of the compounds, the rejection by the membrane can be influenced by other factors. In some instances, arsenic compounds that are 1–2 orders of magnitude smaller than the membrane pore size have been rejected, thus suggesting that besides physical straining, other removal mechanisms may be involved [107]. Two factors that play significant roles in arsenic rejection by a membrane are shape and chemical characteristics of the arsenic compounds. Removal of arsenic compounds on membranes can also be accomplished through repulsion or surface adsorption depending on the charge and hydrophobicity of the membrane and the feed water [109]. The achievement of removal efficiency of 97% for As(V) and 92% for As(III) in a single pass by reverse osmosis has been reported [107].

#### 4.3.1. Arsenic Removal by Membrane Distillation

As a non-isothermal membrane separation process, membrane distillation (MD) utilizes a microporous hydrophobic membrane with pore size ranging from 0.01 μm to 1 μm [97]. For effective operation, only vapor and non-condensable gases must be present within the membrane pores and the membrane must not be wet [97,98]. Commercially available hydrophobic micro-porous membranes include polytetrafluoroethylene, polyethylene, polypropylene and polyvinylidenefluoride membranes [98]. The simplest and most economical and efficient of the different kinds of MD is direct contact membrane distillation (DCMD) [97,98]. DCMD directly separates the hot feed and the cold permeate with the aid of the membrane. Up to 100% of arsenic has been removed from contaminated groundwater with DCMD [97,98].

#### 4.3.2. Membrane and Adsorption Process Hybrid

New technologies with different processes have also been adopted for further improvement of arsenic removal [2]. Membrane technologies along with low cost adsorptive media have been shown to be effective for the removal of arsenic from water [110]. This has led to the production of a cheap, easy to operate filter system for delivering safe and arsenic-free drinking water. This system consists of three basic components: an organic membrane, a tank/drum in which the membrane is inserted and an adsorptive cartridge made of industrial waste products [110].

#### *4.4. Adsorptive Processes*

Adsorptive processes include ion exchange [94,95], iron oxide/hydroxide coated sand [89,90,111,112], iron-hydroxide coated alumina [91], granular ferric hydroxide (GFH) [113,114] and natural iron ores [115]. A desirable feature of the available adsorbents for developing countries is that they should be cheap, readily available and effective [116].

The basis of adsorption methods is the ability to achieve the attraction of soluble arsenic species to the surface of a suitable sorbent. These methods are cost-effective and have attracted a considerable interest for the development of adsorbents that are capable of cheap and efficient arsenic removal from drinking water. The two most commonly used sorbents are based on: (a) iron compounds, including akaganeite (β-FeOOH), amorphous hydrous ferric oxide (FeOOH), goethite (α-FeOOH) and poorly crystalline hydrous ferric oxide [111], and (b) aluminum compounds, including activated alumina γ-Al2O3 and gibbsite Al(OH)3 [117].

Among other sorbents considered are carbon from coconut husks [118], carbon from fly ash [119,120], hybrid polymeric materials [121], red mud [122,123], titanium dioxide [124,125], manganese dioxide [126,127], orange peel [128] and fungal biomass-based bio-sorbent [86,129].

#### 4.4.1. Iron Oxides/Oxyhydroxides Based Adsorbents

The effectiveness of various iron compounds for metal ion removal has been demonstrated in a number of studies [35,130]. Those found to be effective sorbents for arsenic removal from aqueous solutions include β-FeOOH, FeOOH, α-FeOOH and poorly crystalline hydrous ferric oxide [80,90,111,112]. Other iron oxides/oxyhydroxides-based sorbents that have been considered include Cerium (IV)-doped iron oxide adsorbent [115], GFH [113,114], iron-hydroxide coated alumina [91], iron oxide-coated polymeric minerals [100,131], iron oxide-coated sand [132], magnetic iron oxide/activated carbon composite [132], iron oxide-coated cement [96], magnetically modified zeolite [90], natural iron ores [115], silica-containing iron(III) oxide [133] and iron containing waste materials such as fly ash and red mud [120].

The adsorption of arsenic by hydrous ferric oxide (FeOOH), particularly akaganeite, ferrihydrite, goethite and GFH, has been demonstrated to be very strong and effective [114]. The breakthrough behavior of a GFH fixed bed filter was investigated for arsenic removal and maximum uptakes of 28.9 mg/g and 42.7 mg/g at pH 7 were achieved with initial low and high concentrations of 0–1 mg/L and 1–8 mg/L, respectively [114]. The use of goethite for arsenic removal also achieved adsorption capacity of 25 mg/g [112]. The ability to use goethite to achieve a high arsenic removal rate was suggested in another study [134]. The achievement of higher arsenic uptakes of 65 mg/g at pH 3.5 and 22 ◦C [90] and of 120 mg/g at a pH range of 4.5–7 and 25 ◦C [111] has been successfully demonstrated with the use of a synthetic β-FeOOH.

The iron oxyhydroxides used in these studies have particle sizes in the nanometer range. However, when used in sorption columns, nanoparticles can be problematic due to the difficulty in achieving solid/liquid separation [111]. It is therefore a major challenge to find a natural adsorbent that contains nanoparticles of iron oxides that can remove arsenic effectively and can also be used in column filtration technology.

The use of iron oxide coated sand (IOCS) as an adsorbent for removal of arsenic and iron has been investigated by UNESCO-IHE (The Netherlands) for 20 or more years. It is obtained as a by-product of the iron removal process within the Dutch groundwater treatment plants. The investigation of the use of IOCS obtained from different groundwater treatment plants in the Netherlands demonstrates that high to very high arsenic removal efficiencies were achieved for As(III) and As(V) depending on the iron and manganese content in their coatings. Arsenic removal efficiency of 100 percent was achieved with IOCS which contained 353.8 mg Fe/g IOCS and 17.2 mg Mn/g [135].

#### 4.4.2. Application of Nanoparticles for Arsenic Removal from Water

Various nanomaterials have been employed for water treatment [136]. Their reactivities and high surface areas have attracted a lot of interest in their consideration as novel adsorbents for removal of heavy metals and arsenic [137]. Maiti et al. [138] have reviewed the use of nanomaterials for arsenic removal from water and some of the key nanomaterial technologies are summarized in this section. The implementation of these technologies has been limited by the costs of production and the difficulty in recovery and recycling of the nanoparticles. Of these, iron- and titanium-based nanoparticles are most commonly used for removal of arsenic [139–143].

The nano-adsorbents that have been considered for arsenic removal include nanoparticles of zero-valent iron (nZVI) [144,145], iron oxide nanoparticles ∝-Fe2O3, Fe 3O4 and ∝- FeOOH [139,146,147]. The arsenic removal capacity achieved with the iron oxide nanoparticles is influenced by the oxidation state of iron [143].

The oxidation of As(III) to As(V) is supported by the metal core and a thin layer of amorphous iron (oxy)hydroxide present in nanoparticles of nZVI [148]. Although nZVI is a useful adsorbent for arsenic removal, the production of toxic solid wastes from its synthesis is a main disadvantage [149].

Due to their high surface-to-volume ratios [75], iron oxide nanoparticles are 5–10 times more effective for arsenic removal than with the use of micron-sized particles [139]. Ultrafine hematite α-Fe2O3 nanoparticles were successfully synthesized and used for treatment of laboratory-prepared and arsenic contaminated natural water [150]. Rapid removal of As(III) and As(V) was achieved, enabling removal of 74% of As(III) within 30 min. The BET (Brunauer–Emmett–Teller) specific surface area was 162 m2/g and the average particle diameter was 5.0 nm. The achieved adsorption capacity for As(III) was 95 mg/g, while for As(V) it was much lower at 47 mg/g.

The performance of magnetite nanoparticles which have a BET surface area of 69.4 m2/g and a mean particle diameter of 20 nm has also been investigated for arsenic removal [151]. At pH > 7, the adsorption of As(V) rapidly decreases, while a more consistent adsorption was achieved for As(III) at pH 2–9. The achieved adsorption capacity for As(III) was 8.0 mg/g and was 8.8 mg/g for As(V). It was found that the presence of phosphate interfered with arsenic removal [151] and its removal prior to the treatment is recommended.

The removal of As(V) from water was investigated with a nanocomposite of silica and goethite [146]. The synthesized silica nanoparticles had particle sizes ranging from 150 to 250 nm. The achieved adsorption capacity for the goethite/silica nanocomposite was 17.64 mg/g at pH 3.0. The adsorption of As(V) on the nanocomposite was rapid, reaching equilibrium within 120 min [146].

Arsenic removal and photocatalytic oxidation of As(III) have been investigated in a number of studies with nanocrystalline titanium dioxide (TiO2) [152–155]. The adsorption of As(III) and As(V) reached equilibrium on this nano-adsorbent within four hours and 80% or more of both species were adsorbed [152]. The photocatalytic efficiency of nanocrystalline TiO2 in oxidizing As(III) was also demonstrated, achieving full conversion in the presence of sunlight and dissolved oxygen to As(V) within 25 min [152].

Another titanium-based nano-adsorbent for As(III) which does not require oxidation to As(V) is hydrous titanium dioxide nanoparticles (TiO2 × H2O) [156]. High maximum adsorption capacities have been achieved with TiO2 nanoparticles for As(III). An adsorption capacity of 83 mg/g was achieved at a near neutral pH, while at pH 9 it was higher at 96 mg/g [157]. The titanium dioxide nanoparticles provided an effective, low-cost and single-step process for arsenic removal from contaminated water. Nevertheless, due to their particle sizes in the nanometer range, some care and extreme caution must be taken to prevent the possible dispersion of these nanoparticles into the environment [157]. This may be prevented or avoided by either granulation of these nanoparticles into micron-sized particles or loading onto very porous host materials [157].

The reported removal efficiencies achieved by some arsenic removal processes are summarized in Table 3. Evidently, the removal processes based on the use of Fe2O3 filter,

As(III) oxidation by (OCl−) and Fe precipitation, enhanced coagulation/filtration with ferric chloride, iron doped activated carbon, hybrid activated alumina, iron based sorbents, layered double hydroxide (LDH), modified zeolites and laterite and limonite achieved removal efficiencies ≥95%, thus demonstrating that processes involving oxidation and filtration, precipitation and adsorption are effective for removal of arsenic from contaminated water. The chosen process therefore depends on the simplicity and convenience of use if to be adopted for providing drinking water for human consumption.


**Table 3.** Treatment efficiencies reported for some arsenic removal processes.

Adapted and modified from [75].

#### **5. Household Filter for Arsenic Removal for Drinking Water**

A major and impactful outcome from the various reported approaches for treatment of groundwater for removal of arsenic is their adoption in designing potable household filters that have been successfully used for treatment of groundwater as a source of domestic drinking water, particularly in under-developed countries. Several designs of arsenic filters have been developed, proposed and evaluated. Yet, there are still emerging new developments in this area.

One of the most successful domestic treatment units is the SONO filter which is a two-bucket system [158]. The composition of the upper bucket includes a composite iron layer (4–5 cm thick) of a mixture of metal iron and iron hydroxides. This layer is then covered with sand layers. Also, sand and charcoal layers are included in the lower bucket for removal of the iron hydroxides and residual organic matter. Figure 5 shows that the arsenic contaminated groundwater passes through coarse sand, composite iron matrix (CIM), brick chips and wood charcoal [158].

**Figure 5.** SONO Filter. (**a**) Schematic and (**b**) in domestic use [158]. Reproduced with permission from Taylor & Francis.

The SONO filter which is manufactured in Bangladesh utilizes a composite iron matrix (CIM) to remove arsenic by complexation reactions on the metal surface as follows [158]:

$$\text{FeOH} + \text{H}\_2\text{AsO}\_4^- \rightarrow \text{FeHAsO}\_4^- + \text{H}\_2\text{O} \text{ and}$$

$$\text{FeOH} + \text{HAsO}\_4^{2-} \rightarrow \text{FeAsO}\_4^{2-} + \text{H}\_2\text{O}$$

These reactions occur without the need for any pre- or post-chemical treatment. Also no regeneration is required and no toxic wastes are produced [158]. The resulting spent material from the arsenic removal process is a solid self-contained iron-arsenate cement which is non-toxic and does not leach out when in contact with rainwater.

The arsenic concentrations in contaminated groundwater used in one study ranged from 5–4000 μg/L, but these were reduced to 3–30 μg/L after treatment [158]. The treated waters from the filter were found to meet the limits of 10 μg/L and 50 μg/L set by WHO and the Bangladeshi government, respectively. Also, the filter is relatively cheap costing about \$40 for over five years operation and capable of producing 20–30 L/hour to support one to two families drinking and cooking needs [158]. The approval for use of the filter was granted by government and at the early stages about 30,000 of these filters were provided throughout Bangladesh. Its use subsequently spread to India, Nepal and Pakistan. The National Academy of Engineering has recognized the innovation of the SONO filter for arsenic removal by awarding it the Grainger Challenge Prize for sustainability due to "its affordability, reliability, ease of maintenance, social acceptability, and environmental friendliness [158]".

With regards to ongoing maintenance of SONO filters, an important requirement is the need to flush each bucket with 5 L of hot water to eliminate pathogenic bacteria and minimize coliform counts. The filter is expected to last for five years, but if the flow rate decreases, it can be improved by removing the sand layers for washing and reuse or replacement with new sand.

Another commonly used arsenic filter for treatment of contaminated water for arsenic removal is the KanchanTM Arsenic Filter (KAF) [159]. This filter was developed in 2003 through a collaborative effort between the Massachusetts Institute of Technology (MIT) and Environment and Public Health Organization on the mitigation of arsenic contamination of groundwater [160].

Figure 6 shows that the components of a KAF include a plastic container, PVC pipe, diffuser basin, brick chips, iron nails, fine sand, coarse sand and gravel [160]. The arsenic contaminated groundwater is poured into the diffuser basin where it comes into contact with the brick particles and iron nails. Upon contact with water and air, the iron nails rust, consequently producing ferric hydroxide particles that quickly adsorb arsenic from the water. The adsorption process is repeated when new iron surfaces are exposed as the outer surfaces scale off. These continuous processes result in the retention of arsenic on the filter components and produce drinkable filtered water. In addition, the KAF is capable of removing pathogens from the water.

**Figure 6.** Components of a Kanchan Arsenic Filter [160]. Reproduced with permission from Taylor & Francis.

Factors influencing the performance of a KAF include arsenic concentration, duration of use, filter maintenance, flow rate, monitoring and handling and other water components (such as water hardness, iron, chloride and phosphate concentrations) [161]. The effectiveness of KAFs in reducing the concentration of arsenic in groundwater to less than 50 μg/L has been demonstrated [160]. But the effectiveness of the filter can be affected occasionally by the unexpected and unpredictable variability of groundwater conditions, arsenic concentration and climatic conditions.

One of the more recently developed filters is the Pakistan Arsenic Filter (PAF) [162]. This unit employs three different forms of iron (mesh, nails and slag). The iron is held down in a plastic bucket (25 L) with diffuser plates. A tap is attached to the bucket to enable the adjustment of water level and for taking samples at the completion of the treatment, as shown in Figure 7.

**Figure 7.** Components of a Pakistan Arsenic Filter (PAF) and its use for arsenic removal from groundwater [162]. Reproduced with permission.

Figure 7 shows the specific components of a PAF. The round gravel layer overlaid by sand layer can be seen at the bottom. The unit used 1 kg of iron (mesh, nails and slag) contained within the diffuser plate in the bucket and the rate of water flow through the filter is 30 L/min which decreased over an 8-week trial period to 20 L/min [162]. The arsenic contaminated groundwater goes through a diffuser plate and then passes through the sand and gravel. Treated water is transferred through a plastic pipe via a tap, as shown in Figure 7. The highest efficiency achieved for arsenic removal was with the use of iron mesh due to its larger surface area which generates more ferric hydroxide. The PAF was produced at a cost of US\$ 5 which is affordable and accessible to more people [162]. However, it is only applicable to contaminated water which contains 50 to 100 μg/L arsenic, but it was suggested that the unit can be used in series to achieve improved arsenic removal.

Besides the few examples of arsenic filters highlighted in this paper, there are many other filters used or proposed for removal of arsenic from groundwater for domestic consumption. We have decided to describe only the selected few that have had significant impact for domestic provision of suitable drinking water. Also, it is important to note that there is ongoing research to develop simpler and easier to use filter systems.

#### **6. Conclusions and Future Considerations**

This review has demonstrated that arsenic contamination of groundwater impacts the availability of safe and good quality water for domestic use (drinking and cooking) and agriculture in under-developed countries such as Bangladesh, India, Pakistan and Nepal, as well as some parts of more developed countries such as the USA. This can have dire consequences on human health, agriculture, economic and social well-being. The need for a good understanding of the chemistry of arsenic and its fate in the environment has also been demonstrated to be very important for dealing with or minimising the consequences of groundwater contamination by arsenic. The case study about Bangladesh provided a sharp focus on the dire consequences and impact of groundwater contamination by arsenic with serious health, agricultural, social and economic impacts. Furthermore, it highlights that when groundwater contamination by arsenic is identified, rapid efforts need to be directed to reducing/minimizing the associated impacts by adopting and utilising relevant treatment and removal strategies, as well as seeking alternatives for drinking water sources where possible. Socio-economic factors will also play a significant role in the adoption of safer drinking water options. A recent study in Bangladesh illustrated the need for public education combined with the provision of alternative drinking water sources to overcome attitudes and the reluctance of people to switch to better alternatives [163]. The adoption of many of these strategies in the development of various filter systems for arsenic removal

has had a great impact on the ability of people to access safe drinking water in some parts of the affected countries and communities. Many of the available potable treatment systems are effective in ensuring the removal and minimisation of arsenic in groundwater dependent communities to acceptable local standards. However, there are still issues with the long-term maintenance of these systems, especially in ensuring ongoing effective removal of arsenic, disinfection, filter replacement and sludge disposal. Ongoing education of local communities on how to maintain these systems, including how often the filters should be replaced and disinfection protocols is necessary to ensure safety, particularly where there is no alternative water source to groundwater. There is therefore still a need for ongoing research on developing more robust and highly effective filter systems for arsenic removal from groundwater for use for domestic consumption, with the aim of further reducing capital and operation costs, improving user friendliness, minimizing maintenance requirements and resolving or eliminating the need for sludge management and disposal. Also, portable devices that can be adopted for household and community monitoring of arsenic concentrations in drinking water are now available [164,165]. Das et al. [164] have developed two low-cost field test kits for detection of arsenic in water. These kits are capable of detecting as low as 10 μg/L of total arsenic in groundwater within 7 min. In addition, there are commercially available quick test systems for arsenic detection in water within 12 min [165]. If adopted, these devices will ensure communities can adequately assess their water safety and, thus, minimize the incidences of arsenic poisoning. Furthermore, the use of various nanomaterials will play a significant role in the development of a next generation efficient filter system. In the long-term, the development of a more affordable version of the ArsenicMaster Whole House Arsenic Water Filtration System [166], shown in Figure 8, which is maintenance-free will go a long way to addressing the various issues with existing filters if it can be designed to suit the arsenic levels found in groundwater in the under-developed countries and remote communities at a reasonable capital cost.

**Figure 8.** Components of the ArsenicMaster Whole House Arsenic Water Filtration System and a commercially available version [166]. Reproduced with permission.

**Author Contributions:** Conceptualization, S.B.A. and S.K.; methodology, S.B.A., S.K. and A.F.P.; formal analysis, S.B.A., S.K. and A.F.P.; investigation, S.B.A. and S.K.; resources, S.B.A. and A.F.P.; data curation, S.K and S.B.A.; Writing—Original draft preparation, S.K., S.B.A. and A.F.P.; Writing— Review and editing, S.K., S.B.A. and A.F.P.; visualization, S.B.A.; supervision, S.B.A. and A.F.P.; project administration, S.B.A.; funding acquisition, S.B.A. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

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

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable for a review paper.

**Acknowledgments:** The provision of postgraduate research scholarships and research facilities by Monash University to S.K. is gratefully acknowledged.

**Conflicts of Interest:** The authors declare no conflict of interest and that 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**


### *Article* **The Effect of Superabsorbent Polymers on the Microstructure and Self-Healing Properties of Cementitious-Based Composite Materials**

**Irene A. Kanellopoulou, Ioannis A. Kartsonakis and Costas A. Charitidis \***

School of Chemical Engineering, R-Nano Lab, Laboratory of Advanced, Composite, Nanomaterials and Nanotechnology, National Technical University of Athens, 9 Heroon Polytechniou str., Zografou Campus, 15773 Athens, Greece; ikan@chemeng.ntua.gr (I.A.K.); ikartso@chemeng.ntua.gr (I.A.K.) **\*** Correspondence: charitidis@chemeng.ntua.gr; Tel.: +30-210-772-4046

**Featured Application: Superabsorbent polymers of novel structure have been used in cementitious-based composite materials improving their self-healing behavior by an index of 60%.**

**Abstract:** Cementitious structures have prevailed worldwide and are expected to exhibit further growth in the future. Nevertheless, cement cracking is an issue that needs to be addressed in order to enhance structure durability and sustainability especially when exposed to aggressive environments. The purpose of this work was to examine the impact of the Superabsorbent Polymers (SAPs) incorporation into cementitious composite materials (mortars) with respect to their structure (hybrid structure consisting of organic core—inorganic shell) and evaluate the microstructure and self-healing properties of the obtained mortars. The applied SAPs were tailored to maintain their functionality in the cementitious environment. Control and mortar/SAPs specimens with two different SAPs concentrations (1 and 2% bwoc) were molded and their mechanical properties were determined according to EN 196-1, while their microstructure and self-healing behavior were evaluated via microCT. Compressive strength, a key property for mortars, which often degrades with SAPs incorporation, in this work, practically remained intact for all specimens. This is coherent with the porosity reduction and the narrower range of pore size distribution for the mortar/SAPs specimens as determined via microCT. Moreover, the self-healing behavior of mortar-SAPs specimens was enhanced up to 60% compared to control specimens. Conclusively, the overall SAPs functionality in cementitious-based materials was optimized.

**Keywords:** mechanical properties; microstructure; self-healing; SAP; microCT; cementitious materials; mortar

#### **1. Introduction**

Cementitious materials have been widely used over the years in the construction sector due to the cement abundance, its low cost and excellent durability [1,2]. In 2017 the global cement production came up to 4.1 billion tones. China and India which currently represent two of the most rapidly growing countries worldwide produced 57% and 7% of the global cement production in that year, while 6% of the same production is attributed to Europe [3]. It is estimated that the cement production will double in the decades to come [2,4].

Nevertheless, cement is prone to cracking as a result of both inherent material properties such as low tensile strength and application related factors such as tension inducement during the infrastructure service life thus compromising its integrity and durability [1,5–7]. Aggressive environment conditions (i.e., wide ambient temperature fluctuation, rich presence of ions, pH levels etc.) combined with external loading favor crack propagation often

**Citation:** Kanellopoulou, I.A.; Kartsonakis, I.A.; Charitidis, C.A. The Effect of Superabsorbent Polymers on the Microstructure and Self-Healing Properties of Cementitious-Based Composite Materials. *Appl. Sci.* **2021**, *11*, 700. https://doi.org/10.3390/ app11020700

Received: 30 November 2020 Accepted: 10 January 2021 Published: 13 January 2021

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

**Copyright:** © 2021 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/).

leading to the formation of an interconnected crack network that allows corrosive factors to penetrate the structure and assault the reinforcement thus leading to its deterioration. This raises huge safety issues and makes maintenance and repair high priority concerns [6–8].

Even though autogenous crack self-healing in cementitious materials has been known for centuries, its effectiveness highly depends on numerous factors, namely the crack width and age, water abundance upon the crack formation, environment conditions such as pH and presence of ions etc. [6,9–11]. Conventional retrofitting/repair methods applied up to now mainly consist of textile reinforced mortar/concrete (TRM/TRC), the textiles used mainly being based on carbon, glass and aramid fibers, polymer crack injection and polymer modified concrete [1,9,10]. These methods have proven to be effective but in several cases, the cost is prohibitive and/or these methods are difficult or even impossible to apply because of the location of the damaged spot on the infrastructure [6,9].

As a result, internal curing promoting methods have gained a lot interest in the last two decades [1,9]. These methods focus on (i) maintaining continuous water provision and thus continuous cement hydration and (ii) restraining cement self-desiccation [1]. Over the years, the incorporation of several internal curing agents in cementitious materials such as Light Weight Aggregates (LWA) [1,2,5,9,12,13], Superabsorbent Polymers (SAPs) [9,10,14–26], Rice Husk Ash (RHA) [1,5,27], bottom ash [1,5,28], fly ash [1,2,5,9,13,29], cenospheres [1,5,30], crushed returned concrete fine aggregates (CCA) [1,13] and wood fibers [1,13] has been tested. In this application field, SAPs seem to have very promising results and gain more research interest [10].

The SAPs are 3-D polymer networks that due to their hydrophilic nature absorb huge amounts of water (even thousands of times their own dry weight), while due to the network crosslinking they retain their structure and are not dissolved [1,14,15,19,22,24,31]. It is confirmed that their total water absorption level is inversely dependent on environment related factors such as the presence of ions and pH values [10,18,22,24].

Due to their properties, SAPs are used in a vast variety of applications such as hygienic products, agriculture, drug delivery systems, sealing, pharmaceuticals, biomedical applications, tissue engineering, biosensors and the construction field [32]. In the construction field, attention has been drawn to different strategies to obtain coatings with debonding properties [33]. One of these strategies is to incorporate SAPs into an intermediate primer layer between the substrate and the top coatings. The trigger mechanism relies on the fact that with a pH variation, the SAPs can enhance their shape due to the water absorption resulting in the reduction of the attachment between the primer layer and both the top coating and substrate, enabling the detachment of the top coating from the corresponding substrate. Moreover, in cementitious pastes a large number of ions are present (Ca2−, K+, Na+, SO4 <sup>2</sup>−, OH<sup>−</sup> etc.) and pH of the mix water ranges between 11 and 13.4 [10,15,22,24,34].

The incorporation of SAPs in cementitious materials initially results in water absorbance during the cement paste mixing procedure and act as water reservoirs that will make water available during cement curing and hardening. In this way, SAPs contribute to maintaining higher levels of relative humidity thus mitigating early age shrinkage and at the same time favoring extended hydrating reactions which lead to denser cement microstructure by reducing the capillary pores in the cementitious matrix which leads to improved structure strength [9,10,15–17,20,23,24]. Furthermore, upon crack formation and in the presence of water, SAPs reabsorb water on a separate event, swell and allow an immediate crack self-healing effect while the promotion of additional hydrating reactions of remaining unhydrated cementitious phases provide a crack self-healing effect, thus acting as internal curing agents [9–11,14–17,23]. On the other hand, when water is released from SAPs particles in the cement matrix, SAPs deswell forming voids around them in the scale of macropores. These macropores are likely to act as strain inducers and be accounted for any strength loss detected in the final structure, a competitive effect to the internal curing promotion that was previously described [9,10,15,17,21,31,35].

It must be clarified that the overall SAPs behavior in cementitious materials highly depends on the nature of SAPs used and their characteristics such as structure, absorption/desorption behavior, morphology (particles shape, size and size distribution) and dosage in the cementitious matrix, as well as water to cement (w/c) ratio and the incorporation procedure adopted [9,15,17,31,35]. In most cases SAPs absorption/desorption has been examined in extracted or synthetic solutions. Nevertheless, there have also been some studies on the in-situ evaluation of this behavior when SAPs are incorporated in a porous cementitious matrix. The results from these studies revealed that SAPs desorption when in contact with a porous, cementitious material, is effected by the bonding between SAPs particles and the cementitious matrix and is governed by diffusion between SAPs particles and capillary sorption in the matrix [36–38]. Consequently, it is imperative to investigate the correlation between SAPs used and the response of the cementitious system with respect to its mechanical properties and its microstructure.

The vast majority of commercial SAPs are copolymeric networks based on acrylic acid or acrylamide that may or may not have been partially neutralized [10,11,14,15,17,18,20–23,31,35]. Because of the diversity of the parameters that have to be satisfied in the construction applications to enhance their internal curing action, SAP particles should preferentially have a homogeneous spherical shape, size in the submicron area so that the voids left behind after their deswelling are smaller and don't affect the structure mechanical properties and chemical affinity to the cement matrix so that they are more easily dispersed homogeneously in the cement paste [23].

The aim of this work was to examine the impact of the incorporation of tailored SAPs with respect to their structure (hybrid structure consisting of organic core—inorganic shell) on mortars mechanical strength in terms of flexural and compressive strength, microstructure and self-healing behavior. The synthesis and characterization of the incorporated SAPs have been thoroughly discussed in previous authors' work. More specifically, the SAP particles used were spherical in the submicron range based on poly (methacrylic acid) crosslinked with ethylene glycol dimethacrylate which were synthesized via radical polymerization and later encapsulated with CaO-SiO2 inorganic shell via the sol-gel method [25,39]. The incorporation of SAPs in mortars was conducted in two different dosages, 1% and 2% by weight of cement (bwoc). The results obtained in this work revealed that the flexural strength improved by 3%, while the compressive strength remained practically intact for the mortar/SAP composite materials compared to the control specimens. Moreover, the total and closed porosity of the mortar/SAPs specimens were reduced by about 0.5% and 2.5% for mortar-SAPs-1 and mortar-SAPs-2, respectively, while self-healing behavior was enhanced for both SAPs concentrations (in the case of mortar-SAPs-1 by 60% and in the case of mortar-SAPs-2 by 10% compared with mortar-reference specimens).

The added value of this work resides in the optimization of the SAPs functionality in cementitious-based materials and the improvement of the cementitious materials selfhealing properties due to tailored SAPs structure, which are easy to fabricate via the combination of the sol-gel process, radical polymerization and the coprecipitation method. Finally, a new approach for the quantitative evaluation of mortars self-healing behavior was proposed.

#### **2. Materials and Methods**

#### *2.1. Materials*

Sand grade in accordance to CEN, EN 196-1 standard, cement CEM I 52.5 N and in-house synthesized SAPs were used to manufacture conventional mortar specimens. As mentioned earlier, the SAPs used were synthesized according to previous authors' work [39]. In Figure 1 the novel SAPs structure which consists of a hybrid organic core of poly(methacrylic acid) crosslinked with ethylene glycol dimethacrylate encapsulated with a composite inorganic shell of silicon-calcium oxide, P(MAA-co-EGDMA)@CaO-SiO2 (Figure 1a,b) is shown. Their size ranges from 190 to 320 nm (Figure 1c) while their maximum water absorbance ratio in cement slurry filtrate is determined 1100% their initial

dry weight. Before their incorporation in mortars, SAPs are ground to fine powder form, in order to dismantle agglomerates and enhance their performance.

**Figure 1.** Polymerization reaction to form the organic core P(MAA-co-EGDMA (**a**), schematic representation of organic core inorganic shell (**b**), SEM image of the synthesized SAPs/P(MAA-co-EGDMA)@CaO-SiO2 (**c**), revealing the homogenous spherical morphology of SAPs particles in the nanoscale.

#### *2.2. Methods*

#### 2.2.1. Manufacture of Mortar/SAPs Specimens

Two specimen series of mortar/SAPs composites were manufactured, prismatic and cylindrical. Prismatic test specimens were fabricated according to the EN 196-1 [40] using a Teflon mold, while cylindrical specimens were using a medical disposable syringe [41–43]. The specimens were cast from a batch of mortar paste with a water/cement ratio (w/c) of 0.50. In accordance to literature review and previous authors' work on cement/SAPs composites [7,10,11,15,16,20,21,23,31,35,39], SAPs were incorporated in mortar in two different dosages, 1 and 2% by weight of cement (bwoc). The SAPs incorporation in the mortar paste was accomplished in succession to the sand fraction and the obtained mortar paste was mixed until the visually monitored dispersion was acceptable. In all cases mixing time needed was more than 30 s. A mortar mixer (MATEST-E094) was used for the dry mechanical mixing of the mixture components and a jolting apparatus (MATEST-E130) to compact the specimens. The molded samples were stored in the moist air room for 24 h and after demolding they were submerged horizontally in a suitable container at 20 ◦C for 28 days. The composition of the specimens is shown in Table 1.

**Table 1.** Composition of the mortar specimens.


2.2.2. Mechanical Properties of Mortar/SAPs Specimens

The flexural strength of mortar/SAPs composites was evaluated using prism specimens with dimensions 40 mm × 40 mm × 160 mm according to EN 196-1 using a universal testing machine of capacity 300 KN (Instron 300DX-B1-C4-G6C) [40]. For each SAPs dosage as well as for control samples, triplicates were manufactured, demolded and tested after 28 days curing in water. The prism halves from the flexural strength tests were used for the compressive strength tests, which were also performed according to EN 196-1.

#### 2.2.3. Microstructure of Mortar/SAPs Specimens

X-ray micro computed tomography (microCT) was utilized to evaluate the mortar/SAPs composites microstructure. This technique is based on the correlation between X-rays absorption, material density and atomic number. High-density materials absorb

X-rays more profoundly and produce light grey projection images. On the other hand, low-density materials are visualized as darker projection images. During the scanning process, angular projections of the specimen were acquired and saved. After the angular projections acquisition was completed, their reconstruction followed. The reconstruction process was executed by the "NRecon" visualization program via the implementation of Feldkamp algorithm. As a result, a 3D reconstructed model of the scanned specimen was produced. Moreover, quantitative parameters were determined using densiometry and morphometry evaluation, the latter based on image segmentation (black and white) which was done via a global threshold method, the Otsu method [44].

In this work, the specimens were scanned using SkyScan 1272 X-ray micro-tomograph at the age of 28 days. The specimens' geometry was chosen to be cylindrical; their diameter was 10 mm and their height ranged between 20 and 30 mm. In order to enhance contrast in microCT images and improve grey scale histogram segmentation iodine was utilized as a contrast agent. Therefore, all specimens were treated with a 3% iodine solution in ethanol prior to their scans. More specifically, at the age of 28 days, the cylindrical specimens were submerged in the iodine solution for 48 h and then, they were dried in an oven at 80 ◦C for 24 h. If the specimens were not immediately scanned, they were stored in a desiccator [45]. The acquisition settings of the scans are presented in Table 2. In addition, for each scan the flat field correction was applied.

**Table 2.** Acquisition settings of the X-ray micro-tomography scans.


The reconstruction and the porosity analysis were performed using NRecon (version 1.6.6.0) and CTAnalyzer (version 1.13) softwares, respectively. The selected volumes of interest were 441, 105 and 350 mm<sup>3</sup> for mortar-reference, mortar-SAPs-1% and mortar-SAPs-2%, respectively. Figure 2 shows representative 2D and 3D reconstructed images of the scanned specimens ((a), (b) mortar-reference; (c), (d) mortar-SAPs-1%; (e), (f) mortar-SAPs-2%).

**Figure 2.** 2D and 3D reconstructed images of the scanned specimens of: mortar-reference (**a**,**b**); mortar-SAPs-1% (**c**,**d**); mortar-SAPs-2% (**e**,**f**).

Moreover, the challenging issues of material phase segmentation and their quantitative analysis were addressed via 2D and 3D reconstructed images data, and more specifically via the grey scale intensity thresholds. In particular, even though microCT does not support chemical analysis, differences in brightness are recorded in the form of Grey Scale Histograms (GSH) due to density and atomic mass variations for different materials. As a result, phase segmentation depends on the thresholding method followed in each case [46]. In this work, GSH were obtained during the scan data processing via CTAnalyzer Software. The Grey Scale Values ranged from "0" to "255". The lower grey scale values correspond to black color in reconstructed images and were attributed to the lack of material or pores. On the other hand, the higher values correspond to white and were attributed to unhydrated cement phases and/or SAPs aggregates, while the intermediary grey scale values are depicted as grey and were attributed to different hydrated cement phases [44,46–50]. In this work, taking into consideration that all hydrated cement phases, mainly Calcium Silicate Hydrates(C-S-H) are expected to show peaks at similar grey scale values, a deconvolution procedure was engaged to identify and quantify the different material phases (pores, hydrated cement phases, unhydrated cement phases and SAPs) assuming that the grey scale values distribution for each phase is Gaussian. Then, a fitting model distribution, comprising a set of four (n = 4) Gauss distributions was numerically fitted to the GSH using an algorithm implemented in Software "Magic Plot" (student version 2.5.1).

#### 2.2.4. Self-Healing Evaluation of Mortar/SAPs Specimens

Cementitious structures can be damaged in a variety of ways, the most common being cracking. MicroCT has been utilized as a laboratory-scale method to evaluate their damage extent and interpret healing mechanisms. In the past more conventional methods, such as SEM (Scanning Electron Microscopy), have been used to evaluate crack morphology and mitigation but they can provide insight only on the specimen surface, whereas microCT can provide useful information on the bulk of the specimen and consequently evaluate internal cracks and internal self-healing [46].

In this work, after 28 days of curing the cylindrical specimens that were previously examined via microCT were precracked under compressive load using a hydraulic press. During their compression, the load was applied in a controlled and smooth manner but the specimens were completely split in two halves in all cases. Therefore, prior to this procedure they were wrapped tightly in a polypropylene based film in order to avoid the complete separation of the two halves but instead to form a crack as shown in Figure 3. Afterwards, the specimen circumference and base were coated with an epoxy resin (a mixture of Sinmast J 158 (component A) and Sinmast S2 liquid primer (component B) by Sintecno in a ratio A:B 77:23) in order to secure the two halves together. The top surfaces of the cylinders were untreated, so that the formed crack could interact with healing agents (water). The epoxy resin was cured at ambient temperature for 1 day and then at 60 ◦C for 2 h.

**Figure 3.** Schematic representation of mortar cylinder (**a**), specimen image from microCT CCD camera (**b**), crack inducement in cylindrical specimens via compression load enforcement (120**c**).

In order to estimate their self-healing behavior, the cracked specimens were submerged in water at ambient temperature and they were evaluated at different time slots (0 and 8 days). The self-healing efficiency of mortar/SAPs composites was quantified and visualized via microCT analysis. Prior to microCT evaluation the specimens were dried at 50 ◦C for three days and were stored in a desiccator. The cracked mortar specimens were scanned using the SkyScan 1272 desktop microCT at 25.0 μm pixel resolution with 0.5 mm aluminum filter. The scanned images were reconstructed via NRecon software, while threedimensional evaluation was conducted by CTAnalyzer software following a methodology proposed by Nicole Y.C Yu et al., appropriately adjusted for cracked mortar specimens [51]. The main outcomes discussed in this work were the evolution of connectivity density and the percent object volume versus healing time. These parameters were adopted because they allow the quantitative evaluation of crack healing in 3D (not only in 2D), with respect to crack closure in terms of changes in morphology and density and are embedded in the CTAnalyzer software.

#### **3. Results**

#### *3.1. Mechanical Properties of Mortar/SAPs Specimens*

During the flexural strength test an abrupt, brittle failure was observed in mortars mainly due to deformation localization by the coalescence of narrow microcracks that ultimately lead to macrocracks that expand to the entire specimen [52]. The flexural strength evaluation of the mortar-based composites with SAPs in dosages 1 and 2% bwoc as well as control mortar specimens at the age of 28 days is demonstrated in Figure 4a,b. According to Figure 4b, a slight increment of about 3% can be detected on the flexural strength of the mortar-based composites that contain SAPs in dosages 1% and 2% bwoc, in comparison to the control specimens which practically reveals that the flexural strength remains intact after the SAPs incorporation in both concentrations.

**Figure 4.** Maximum values of loads for mortar specimens at 28 days of age (**a**), flexural strength of mortar specimens at 28 days of age (**b**), optical image of specimen after flexural test (**c**).

On the other hand, the compressive strength of the mortar based composite materials with the SAPs incorporation for the same mortar age (28 days) and w/c ratio (0.5) is depicted in Figure 5a,b. More specifically, Figure 5a depicts the load applied during the test as a function of the head displacement and Figure 5b shows the calculated compressive strength of the mortar based composite materials with the SAPs incorporation for the same mortar age (28 days) and w/c ratio (0.5). As shown in Figure 5b the compressive strength of the mortar-based composite materials practically remains intact for both SAPs dosages (1% and 2% bwoc) in comparison with the control materials.

**Figure 5.** Maximum values of loads for mortar specimens at 28 days of age (**a**), compressive strength of mortar specimens at 28 days of age (**b**), optical image of specimen after compressive test (**c**).

Figure 6 reveals the relationship between the flexural and compressive strength of the mortar-based control and composite materials using the corresponding average values at the age of 28 days. It can be observed that a direct relationship (R2 = 0.97818) between them exists. Similar behaviors have been reported by other research works for reference mortars [53,54].

**Figure 6.** Relationship between compressive and flexural strength of mortar-based materials.

#### *3.2. Microstructure of Mortar/SAPs Specimens*

#### 3.2.1. Microstructure Analysis

In this work, the effect of SAPs incorporation on the mortar microstructure in means of porosity and phase segmentation was examined via microCT. In cementitious materials, porosity affects greatly key properties such as mechanical and transport properties [46]. Moreover, the dispersion quality of SAPs in the cementitious matrix can be evaluated through porosity determination. The porosity of cementitious materials is usually divided into gel pores (ranging from a few nanometers to 0.2 μm), capillary pores (ranging from

0.2 to 10 μm), and air voids (above 10 μm) [50,55,56]. In this work, the pore analysis from micro-CT applies only to the pores with diameters larger than 9 μm. The specimen size dictates the minimum distance that can be reached from the X-ray source during scanning and consequently the accuracy of the method. Therefore, the pore analysis from micro-CT scans includes partially capillary pores and the air voids. Air voids are mainly of interest in this work, as in literature, it is reported that the incorporation of SAPs in cementitious matrixes creates air voids around them as a result of the water absorption/desorption by them combined with their poor dispersion in cement and therefore the formation of large SAPs agglomerates [9,10,15,21,31,35,38]. The statistical pore analysis of mortar specimens and the corresponding pore size distribution are exhibited in Table 3 and Figure 7, respectively. The statistical pore analysis parameters calculated are delineated as follows.

**Table 3.** Statistical pore analysis of mortar specimens.


**Figure 7.** Pore size distribution in mortar-reference, mortar-SAPs-1 and mortar-SAPs-2.


#### 3.2.2. Image Segmentation and Phase Identification

There are a number of different approaches on the matter of image segmentation and phase identification via microCT Analysis. One that stands out as it is widely used is the global thresholding method in GSH. The GSH with the Gaussian deconvolution procedure performed for the mortars examined in this work, are given in Figure 8.

**Figure 8.** Grey Scale Histogram with the Gaussian deconvolution fit for mortar-reference (**a**), mortar-SAPs-1 (**b**), mortar-SAPs-2 (**c**).

The deconvolution of the initial GSH as described earlier, produced four distinct curves in each case, curves 1 to 4 and the corresponding peaks (Figure 8). Curve 1—Peak 1 were attributed to pores, curve 2 and curve 3 were assigned to hydrated cement phases and curve 4 was ascribed to unhydrated cement phases and SAPs. We observed that for all specimens the main peak in the GSH was deconvoluted in two separate peaks which were attributed to different hydrated products probably owed to the main hydrated cement products which are Calcium Silicate Hydrates, C-S-H and Calcium Hydroxide, Ca(OH)2.

Then, using the GSH data, the quantitative determination of the different phases identified in the mortar specimens was performed and the corresponding results were tabulated in Table 4. Taking into consideration the data in Table 4, it is observed that in the case of mortar-SAPs-1 the unhydrated cement products together with the incorporated SAPs (1% bwoc) represented only the 5% of the total material when the corresponding values for both mortar-reference and mortar-SAPs-2 were 36%. This indicates that the progress of the hydration reactions for this material was remarkably enhanced compared to mortar-reference and mortar-SAPs-2, as a result of SAPs incorporation in the cementitious matrix. This is attributed to more effective SAPs dispersion in the cementitious matrix and therefore enhancement of their functionality to extend hydration reactions during cement curing, thus promoting this as the optimal SAPs concentration in mortars in respect to microstructure evaluation.

**Table 4.** Quantitative determination of the different cement phases identified in the mortars.


#### 3.2.3. Self-Healing Evaluation

In Figure 9 the crack surface and the larger voids in respect to cracks and large holes in the bulk of the specimens immediately after they were cracked (0 days) and after 8 days of healing treatment, are shown for the mortar specimens examined in this work.

**Figure 9.** Initial microCT images depicting voids (in respect to both large pores and cracks) in red color for mortarreference (**a1**), mortar-SAPs-1 (**b1**) and mortar-SAPs-2 (**c1**) and corresponding images after 8 days of healing treatment for mortar-reference (**a2**), mortar-SAPs-1 (**b2**) and mortar-SAPs-2 (**c2**).

MicroCT imaging allows both to visualize the self-healing process, but also to quantitatively analyze it in terms of changes in morphology and density using methods and functions embedded in CTAnalyzer software properly adjusted for mortar specimens. The methodology followed in this work is described as follows.

The region surrounding a crack is rich in products with a vast variety of thicknesses ranging from thick intact mortar to fresh self-healing products that can be thinner or thicker structures. Binarizing or segmenting the structures in this region can therefore be compromised. An effective solution, which has the effect of artificially diminishing the attenuation of thin structures, is the method of adaptive thresholding in CTAnalyzer custom processing procedure. Then, the (Region of Interest) ROI shrink-wrap and stretch over holes functions were performed. As a result, the (Volume of Interest) VOI was wrapped around the boundary of complex and porous objects such as thin self-healing products and the wrapped VOI was prevented from penetrating into the porous spaces of the object but instead only the complex outer margins were marked. Figures 10 and 11 depict the resulting banalization-segmentation images after applying the adaptive thresholding method and shrink-wrap function for mortar-reference, mortar-SAPs-1 and mortar-SAPs-2 specimens at 0 and 8 days of healing treatment, accordingly.

**Figure 10.** Initial microCT images for mortar-reference (**a1**), mortar-SAPs-1 (**b1**) and mortar-SAPs-2 (**c1**), resulting images for mortar-reference (**a2**), mortar-SAPs-1 (**b2**) and mortar-SAPs-2 (**c2**) and ROI images for mortar-reference (**a3**), mortar-SAPs-1 (**b3**) and mortar-SAPs-2 (**c3**) after adaptive thresholding method and shrink-wrap function are applied before healing treatment.

**Figure 11.** Initial microCT images for mortar-reference (**a1**), mortar-SAPs-1 (**b1**) and mortar-SAPs-2 (**c1**), resulting images for mortar-reference (**a2**), mortar-SAPs-1 (**b2**) and mortar-SAPs-2 (**c2**) and ROI images for mortar-reference (**a3**), mortar-SAPs-1 (**b3**) and mortar-SAPs-2 (**c3**) after adaptive thresholding method and shrink-wrap function are applied after 8 days of healing treatment.

Then, a full 3D analysis was run. The morphometric parameters calculated to quantitatively evaluate self-healing of the cracked mortar specimens were (a) percent object volume (%) and (b) connectivity density (mm−3) [51,57,58]. The first parameter shows the specimen volume variations, while the second one is sensitive to structure complexity changes versus healing time. The corresponding results are shown in Figure 12.

**Figure 12.** Calculated morphometric parameters to evaluate the self-healing process of mortars. Percent object volume (**a**), connectivity density (**b**). The error bars refer to percent error.

#### **4. Discussion**

The overall effect of SAPs incorporation in cementitious materials highly depends on the SAPs properties. In this study, in house synthesized SAPs were used with tailored properties [39]. Firstly, different SAP particle sizes introduce different effects. For the same SAP concentration and w/c ratio, the larger the SAP particle size, the lower the determined strength was in means of flexural and compressive strength. This is directly correlated with the formation of larger SAP voids left behind in the cement structure when SAPs release the amount of water they absorbed during mixing [21]. The size of the SAPs used in this work laid in the submicron scale as shown in Figure 1c, whereas the particle sizes of SAPs tested by other researchers were several micrometers [8–10,15,18,21,23].

Additionally, SAPs chemical constitution triggers different behaviors in the cementitious matrix. As mentioned earlier, most commercially available SAPs are copolymeric networks based on acrylic acid or acrylamide that may or may not have been partially neutralized [10,11,14,15,17,18,20–23,31,35] which present a low affinity with cement and as a result form large aggregates when incorporated in it. In this work, SAPs that have been tested had a hybrid organic core—inorganic shell structure. The introduction of an inorganic coating on the organic SAP core has been proposed by Kanellopoulou I. et al. (2019), via the encapsulation of the polymeric core with an inorganic CaO-SiO2 shell via solution-gelation technique [31]. Other research groups have investigated the behavior of coated SAPs via the Wurster process but the compensation of the strength reduction achieved was only partial [9].

On the contrary, as indicated by the total porosity reduction and the fact that no mechanical strength degradation was determined with respect to flexural and compressive strength for the mortar/SAPs composite materials, the tailored SAPs in this work presented enhanced chemical affinity with the cementitious matrix. More specifically, due to the enhancement of the chemical affinity between SAPs and cement, large SAP agglomerates were not formed and as a result the size of the SAP voids (porosity) when water was released from their particles were smaller. Additionally, the limited water absorption capacity of the coated SAPs compared to the corresponding capacity of the uncoated particles also favored the size limitation of the SAP voids. The total porosity reduction in both cases of SAPs incorporation in mortars as shown in Table 3 indicates that:

• SAPs were homogeneously dispersed in mortars due to enhanced chemical affinity with the cementitious matrix and

• SAPs promoted hydration reactions in the cement matrix, thus forming denser and consequently more durable structures, since stress inducing points (voids) in the cement matrix were smaller. This is coherent with the mechanical properties behavior discussed later.

The slight flexural strength increment (~3%) of the mortar-SAPs specimens in comparison to the control specimens was evaluated taking into account as corroborating evidence the aforementioned porosity parameters and was attributed to the water absorbed by SAPs during the mixing process, which became available later during cement curing, thus promoting and accelerating hydration reactions. The space surrounding SAP particles was enriched with hydrated cement phases. As a result, mitigation of autogenous shrinkage and early age cracking were observed, while the densification of the cement microstructure led to the enhancement of the mortar/SAPs composites flexural strength [8,52,54,59].

Moreover, compressive strength is considered a key concrete property since it is directly correlated with concrete quality. Nevertheless, it must be clarified that the determined value of each measurement set highly depends on a variety of parameters namely the w/c, additives incorporation and curing conditions, i.e., humidity levels and temperature profiles. Curing conditions can affect drastically the compressive strength of a cementitious material. More specifically, compressive strength degradation has been correlated with low moisture levels during the first day of curing or high temperatures in the initial curing state which is responsible for lower quality hydration products [52]. Furthermore, w/c ratios used in mortars formulation have also been directly correlated with the voids formed in the concrete matrix and consequently with their mechanical strength. More specifically, increased w/c ratios have been known to lead to increased voids in the cementitious matrix and thus to the degradation of mechanical properties [1]. The w/c ratio used in this work (0.5) is high compared to the corresponding ratios examined by other research groups. For example the w/c ratios in cementitious composites with SAPs were 0.30–0.35 in the research of Sun et al. [8], 0.40 and 0.50 in the research of De Belie et al. [9], 0.30, 0.40, 0.50 in the research of Lee et al. [10], 0.35, 0.40 and 0.50 in the research of Farzanian et al. [15] and 0.40 in the research of Kim [54]. Nonetheless, since the effective w/c ratio is not the same as the total w/c ratio, when SAPs are incorporated in mortars, excess of water must be added to counterbalance the amount of water absorbed by SAPs. Considering the dosage of SAPs used in this work (1 and 2% bwoc) the high w/c ratio of 0.50 was chosen. The w/c ratio chosen combined with the curing conditions of the mortar specimens and notably, low moisture during the first day after casting as well as low temperature during the 28 days specimen curing the value of the compressive strength of the control specimens was calculated below 52.5 MPa.

In literature, it is often reported that cementitious materials containing SAPs show poorer compressive performance compared to the corresponding materials without SAPs during all curing periods and for SAPs dosages even lower than those examined in this work, while this behavior became more pronounced for increased SAPs dosages. [1,7,9,10,15,17]. On the contrary, in this work compressive strength was not influenced by the SAPs incorporation in mortars.

The relationship between the determined flexural and compressive strength for the mortar specimens as depicted in Figure 6 was also indicative of the fact that SAPs incorporation in the mortars affected flexural and compressive strength in a similar manner and that their incorporation did not negatively influence the mortar microstructure and hence the mortar strength was not degraded [53,54].

According to the results in Table 3, the percentage of total porosity was reduced by 0.5% and 2.5% in the cases of mortar-SAPs-1 and mortar-SAPs-2, respectively. Moreover, the percentage of open porosity was also reduced in the case of mortar-SAPs-2 compared to mortar-reference by 0.5%. Open porosity represents the pores which are in direct contact with the specimen environment thus allowing humidity and/or other harmful factors (e.g., corrosive factors) to penetrate the bulk of the specimen and cause the material

deterioration. As a result, the open porosity reduction in cementitious materials promotes their mechanical integrity and their sustainability.

Additionally, as shown in Figure 7, the size of the pores in the mortar-reference specimens was greater than that of the mortar-SAPs-1 and mortar-SAPs-2 specimens, while the smallest pores were recorded in the case of mortar-SAPs-2. The pore sizes in the mortarreference specimens showed a wide distribution ranging from 9 to almost 500 μm. On the contrary, the pore sizes in mortar specimens in which SAPs had been incorporated showed a narrower distribution ranging from 9 to 333 μm and from 9 to 171 μm for mortar-SAPs-1 and mortar-SAPs-2, respectively. The more uniform the pore profile the more homogeneous the materials, which is consistent with improved mechanical properties.

Furthermore, the Gaussian deconvolution procedure of the GSH of all the mortar specimens examined in this work, showed that the minimum unhydrated cement phases (5%) were found in the case of the mortar-SAPs-1 specimens, whereas the corresponding phases in mortar-reference (36%) and mortar-SAPs-2 (35%) were almost identical (see Table 4). Firstly, this indicates that SAPs were more homogeneously dispersed in the cementitious matrix for the concentration of 1% bwoc thus allowing them to function as water reservoirs and cement hydration promoting agents and therefore led to denser and more durable cementitious structures. Secondly, it is safe to assume that even greater SAPs concentrations, if needed to render different functionalities to the mortars, will not compromise the mortar behavior.

In mortars, apart from one single crack several other microcracks are formed and distributed in the bulk of the specimens, when they are subjected to compressive loads, which sometimes may not be easy to detect using conventional methods. On the contrary, by running a full 3D analysis via microCT, apart from crack healing, the 3D volume of the specimen was also be thoroughly evaluated. As mentioned earlier, the quantitative analysis of the self-healing progress in pre-cracked mortar specimens took place via microCT Analysis. More specifically, a morphological and an architectural parameter were calculated for all mortar specimens before and after eight days of healing treatment. These parameters were:


In order to estimate the relative self-healing enhancement of mortar/SAPs specimens versus the control specimens, the relative self-healing enhancement index *hi* (%) was calculated [61], based on the results of Connectivity Density since this architecture parameter is very sensitive to texture and complexity changes.

$$h\_i(\%) = \left(1 - \frac{CD\_{i,\mu}}{CD\_{i,0}}\right) \times 100\,,\tag{1}$$

where:

*hi*: relative self-healing enhancement index for specimen, *i CDi*,*n*: value of Connectivity Density for specimen, *i* after *n* days of self-healing treatment *CDi*,0: value of Connectivity Density for specimen, *i* before self-healing treatment (at 0 days)

More specifically, when no self-healing is observed *hi* value is "0", as the final and the initial value of connectivity density are the same. On the other hand, when selfhealing is promoted, *hi* values become higher and approach the value "100", since the

ratio *CDi*,*n*/*CDi*,0 approaches "0" as *CDi*,*<sup>n</sup>* decreases. The corresponding values of *hi* for the mortar/SAPs and control specimens are shown in Table 5 and it is revealed that the self-healing index was 84% and 31% for mortar-SAPs-1 and mortar-SAPs-2, respectively, while it was 20% for mortar-reference specimen. Comparatively, self-healing was enhanced by about 60% and 10% for mortar-SAPs-1 and mortar-SAPs-2, respectively compared to the mortar-reference specimen. These results reveal that healing efficiency was optimized for the specimen mortar-SAPs-1. This is attributed to the more effective dispersion of SAPs in the cementitious matrix for the SAPs concentration 1% bwoc and therefore the enhancement of their functionality as self-healing agents. This conclusion is in agreement with that drawn by the phase identification in mortars via image segmentation obtained by microCT scans.

**Table 5.** Indexes *hi* for the mortar/SAPs specimens and control specimens.


At this point, it must be taken into consideration that the specimen mortar-SAPs-1 was more severely damaged during the artificial inducement of the crack by the application of compressive load, leading to a very high initial value for connectivity density.

Our future goal is to expand this preliminary study on the evaluation of the selfhealing behavior of mortars containing the proposed SAPs for longer treatment durations and attain a more profound insight on the healing mechanisms.

#### **5. Conclusions**

The presented work, took into consideration the overall behavior of the mortar-SAPs composites manufactured and examined. It also comprises the determination of their mechanical properties, their microstructure evaluation, as well as the evaluation of their self-healing behavior yielding a series of conclusions. Even though it is often reported in literature that the incorporation of SAPs in mortars causes degradation in mechanical properties and specifically in compressive strength [1,7,9,10,15,17], within the manuscript it is shown that, the incorporation of tailored SAPs with respect to their structure (hybrid organic core—inorganic shell structure, spherical shape in the submicron scale) did not negatively influence neither the flexural nor the compressive strength of the mortars. This is directly correlated with the microstructure and porosity evaluation of the mortars, which took place via microCT analysis. In particular, the total porosity was reduced by about 0.5% and 2.5% for mortar-SAPs-1 and mortar-SAPs-2, respectively, while the range of the pore size distribution became narrower for both SAPs concentrations compared to the control specimens. As a result, these SAPs enhanced cement hydration reactions when incorporated in the mortars without introducing more stress inducing sites (macropores left behind in the matrix because of the deswelling of SAPs particles during cement curing) and consequently not compromising the mortars strength. Moreover, the Gaussian deconvolution procedure of the GSH of all the mortar specimens examined in this work, showed that the minimum unhydrated cement phases (5%) after 28 days of aging, were found in the case of mortar-SAPs-1, which also revealed the more pronounced self-healing behavior.

Conclusively, the overall SAPs functionality in cementitious-based materials was optimized while, the SAP concentration of 1% bwoc was promoted as the premium one in reference to mortar composite strength, microstructure and self-healing enhancement.

**Author Contributions:** Conceptualization, I.A.K. (Irene A. Kanellopoulou) and I.A.K. (Ioannis A. Kartsonakis); methodology, I.A.K. (Irene A. Kanellopoulou); software, I.A.K. (Irene A. Kanellopoulou) and I.A.K. (Ioannis A. Kartsonakis); validation, I.A.K. (Irene A. Kanellopoulou), I.A.K. (Ioannis A. Kartsonakis) and C.A.C.; formal analysis, I.A.K. (Irene A. Kanellopoulou) and I.A.K. (Ioannis A. Kartsonakis); investigation, I.A.K. (Irene A. Kanellopoulou); resources, I.A.K. (Irene A. Kanellopoulou) and I.A.K. (Ioannis A. Kartsonakis); data curation, I.A.K. (Irene A. Kanellopoulou); writing—original draft preparation, I.A.K. (Irene A. Kanellopoulou) and I.A.K. (Ioannis A. Kartsonakis); writing—review and editing, I.A.K. (Ioannis A. Kartsonakis) and C.A.C.; visualization, I.A.K. (Irene A. Kanellopoulou) and I.A.K. (Ioannis A. Kartsonakis); supervision, C.A.C.; project administration, C.A.C.; funding acquisition, C.A.C. All authors have read and agreed to the published version of the manuscript.

**Funding:** These results are part of the projects that have received funding from the European Unions' HORIZON 2020 research and innovation program under grant agreement no. 685445 (LORCENIS) and under grant agreement no. 814505 (DECOAT).

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

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Data sharing not applicable.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**

