Development of improved low-cost ceramic water filters for viral removal in the Haitian context

Household-based water treatment (HWT) is increasingly being promoted to improve water quality and, therefore, health status in low-income countries. Ceramic water filters (CWFs) are used in many regions as sustainable HWT and have been proven to meet World Health Organization (WHO) microbiological performance targets for bacterial removal (2–4 log); however, the described viral removal efficiencies are insufficient to significantly reduce the associated risk of viral infection. With the objective of improving the viral removal efficiencies of ceramic water filters, new prototypes with different oxide compositions and firing atmospheres have been developed and evaluated. For removal efficiencies human adenoviruses, MS2 bacteriophage and Escherichia coliwere quantified in all prototypes. A new model of CWF that was fired in a reductive atmosphere presented virus and bacteria removal efficiencies greater than 3.0 log and 2.5 log, respectively, which would fulfill the viral targets that are recommended by the WHO. Ceramic characterization of the selected filters, which were fired in a reductive atmosphere, showed that a larger specific surface area than those of control filters and higher fraction of a positive Z-potential fraction are the most likely explanations for this increase in virus removal. doi: 10.2166/washdev.2014.121 om https://iwaponline.com/washdev/article-pdf/5/1/28/385077/washdev0050028.pdf 2019 L. Guerrero-Latorre M. Rusiñol A. Hundesa S. Bofill-Mas R. Girones (corresponding author) Laboratory of Virus Contaminants of Water and Food, Department of Microbiology, Faculty of Biology, University of Barcelona, Barcelona 08028, Spain E-mail: rgirones@ub.edu M. Garcia-Valles S. Martinez Department of Crystallography, Mineralogy and Mineral Deposits, University of Barcelona, Barcelona, Spain O. Joseph Laboratoire de Qualité de l’Eau et de l’Environnement, Université Quisqueya, Port-au-Prince BP 796, Haïti


ABBREVIATIONS
Large numbers of viruses are excreted in human feces and urine into the environment, and, even at low concentrations, they can cause illness when ingested (Leclerc ).Consequently, sewage and fecal contamination, if not well contained and treated is the main source of microbiological risk in water and food.In countries with poor sanitation systems, such as in Haiti, where only 17% of the population has improved sanitation (Onda ), microbiological contamination in water sources is a major cause of morbidity and mortality.
Most potable water-treatment methods are not suitable or available for the majority of rural areas in low-income countries without reliable access to safe drinking water.
For those populations, household water treatments and storage (HWTS) have been promoted to improve water quality and prevent diarrhea because of two important characteristics low cost and the capacity, from some of them, to be produced locally.Among all of the technologies tested at the laboratory and field levels, ceramic water filters   Firing in a reductive atmosphere consists of firing in a traditional kiln with wood up to the desired temperature (950-990 W C), and then sealing the kiln with the last coal inside the firing chamber.By this technique, the oxygen is totally consumed and a reduction atmosphere affects all clay components and as the iron oxide from the clay becomes ferrous oxide, we can observe, how the color within the wall of the components is gray, rather than brown or red.

METHODS AND MATERIALS
After firing, the filters were cooled down and immersed in water over night.The flow rate was measured after filling the wet filters with water, and filters with acceptable flow rates between 1 and 3 L/h were approved for consequent analysis.No colloidal silver solution was added at any step.A modification of the filtering protocol was also tested by adding a 5-cm layer of crushed ceramic sitting inside the pot in the final prototype to increase the removal efficiency.Ceramic gravel was produced with crushed clay from broken filters fired in a reductive atmosphere.After manual crushing, gravel was sieved with a superior 2.0 mm and inferior 0.5 mm sieve size.A 5-cm layer was selected as the maximum depth to maintain an acceptable flow rate.
Finally, two replicates of the final prototype in real scale (1:1) were performed to evaluate the long-term efficiencies and applicability of the best-performing prototypes.The total numbers of filter units analyzed are shown in Table 1.
To characterize the chemical and mineralogical composition and texture of the raw material that was used and the ceramics that were obtained in this work, several assays were performed on a representative sample after homogenizing the material X-ray fluorescence analysis using a Prototypes of the small-scale filters were prepared and tested using fresh water spiked with E. coli, To assess the capability of the ceramic filter prototypes and to eliminate viruses and bacteria, the small filters
The results obtained using the bacterial standards, viral indicators and tested pathogens are a strong indication of the need for using household-based water treatments and improving CWFs that reduce the level of viral contamination in water.
The prototypes of the CWFs that was developed in this study and tested for microbiological efficiency are shown in Table 1.At the same time, the analysis of the chemical composition of the clays used to produce those CWFs indicates that silica is the major component, and the main difference between the three clays is the calcium content: Artibonite clay had the highest percentage (Table 3).
After firing, all prototypes presented acceptable levels of water filtration capabilities and robustness.The main mineralogical differences between the ceramics that were fired in reducing and oxidizing atmospheres were the presence of magnetite or hematite, respectively, as quartz and feldspar were the dominant phases in both samples.
Moreover, we observed that, in oxidizing conditions, the proportion of calcite (CaCO 3 ) in the ceramic was lower and the content of gehlenite (Ca 2 Al(AlSi)O 7 ) was increased because, in this case, the decomposition was more efficient.However, during reductive firing, the atmosphere seems to restrain the decomposition, which is indicated by the remaining values of calcite (Table 4).Pore size distribution studies of the two firing conditions did not show significant differences between ceramics fired in different atmospheres regarding the pore size distribution showing larger dimensions than viral diameters (Figure 1).Nevertheless, significant differences were found regarding the analysis of the surface area of the ceramic filters, which indicated that filters fired in a reductive atmosphere had greater surface areas, with a mean value of 6.65 m 2 /g, than the filters that were fired in an oxidative atmosphere, with a mean surface area of 2.41 m 2 /g (Table 5).Moreover, studies of Z-potential distribution of ceramics comparing both firing condition shows a fraction of the crushed reductive clay being positive, suggesting a potential effect into viral absorption (Figure 2).
The microbiological removal efficiency of the diverse prototypes was tested with a minimum of two filter replicates per prototype, and the results are shown in Table 6.
The filter control, which was produced as a reference model of a non-silver-impregnated filter, presented viral and bacterial removal efficiencies of 0.57 for MS2 and 0.68 for E. coli, results that are equivalent to previously reported studies (Van der Laan ).
The prototype ceramic filters that were amended with     (Table 6).The analysis of the log removal efficiencies among reductive atmosphere filters were not significant (p > 0.05).
These filters achieved the performance requirements for the HWTS protective technologies that were recently established by the WHO ().
Filters fired in a reductive atmosphere with an extra layer of ceramic gravel were selected to test long-term performance in real-size scale and showed consistent removal values after 1,000 L (Figure 3).During the long-term study, logarithm reduction values were between 2.5 and 4 LRV for the tested viruses (HAdV and MS2), but higher variations occurred for E. coli, with a range of 1.5-4 LRV.
Flow rates were recorded at 1.5-3 L/h and were stable throughout the experiment, and the effluents had a slightly increased pH, from seven in the input water up to eight in the output water, as previously recorded (data not shown) (Halem ).
The new models of a CWF made by firing in a reductive atmosphere and without colloidal silver that have been proposed in this paper showed removal efficiencies for viruses and bacteria that were significantly higher  of CWF to increase microbiological effectiveness without increasing costs.This new improved prototype can be easily implemented in CWF factories as firing in a reductive atmosphere will just require a specific kiln that can be built with local materials (traditional bricks, sand and concrete) or even current kiln models can be accommodated to fire in reductive conditions.Moreover, the fact that colloidal silver can be replaced by a layer of ceramic gravel will also reduce environmental and toxicity hazards as well as economic cost (around 1 euro per filter).At the same time, the addition of this layer made of ceramic gravel will entail a step of maintenance by users because regular cleaning of the gravel will be needed.However, this modification will also improve the microbial performance of the filter and highlights the importance of performing hygiene promotion campaigns alongside the implementation of this product toward users.

CONCLUSIONS
worldwide (42% in sub-Saharan Africa) were still without access to improved sources of drinking water in 2010 (Onda ).Estimates of global young-child (<age 5 years) mortality from 2010 suggest that 2 million young children die due to diarrheal disease, and approximately 1.4 million (70%) of these are localized in low-income countries (Bryce ; Black ; Liu ).Recent data on diarrhea burden in low-income countries have reported 685,000 diarrhea deaths related to the deaths attributable to inadequate water and sanitation in 2012 (Prüss-Ustün ).The etiological origins of diarrhea had been traditionally associated with bacteria and protozoa; however, in recent years, improved diagnostic methods have made possible detection of other etiological agents, such as enteric viruses, which altered the distribution of the most common cause of diarrheal illness worldwide toward enteric viruses (e.g.norovirus (NoV), rotavirus, adenovirus) (Huilan ; Parashar ; Kotloff ).

(
CWFs) have shown the highest impact on long-term health due to the high-adherence of users and their capacity to reduce approximately 50% of the diarrhea incidences (Clasen 4, 5; Brown ; Preez ; Hunter ; Levine ).Currently, the most widely available, locally produced model of CWF is based on a design developed in 1981 by the Guatemalan industrial research institute ICAITI (Instituto Centro Americano de Investigación y Tecnología Industrial), and recently updated by The Ceramics Manufacturing Working Group ().The performance of this CWF with and without colloidal silver has been investigated over the years by researchers who have studied its efficiency at reducing waterborne pathogens in the laboratory and in the field.In the literature, the removal efficiency of bacteria by CWF is described to be within the range of 2-4 logs using the standard types of silver-impregnated CWFs (Halem ; Oyanedel-Craver ; The Ceramics Manufacturing Working Group ); however, viral removal efficiencies have been shown to be much lower, between 0.21 and 0.45 log in CWF impregnated or not impregnated with silver described by the Salsali et al. () study, and around 1.5 log reduction in the Brown & Sobsey () study with no significant differences between impregnated and nonimpregnated filters.On the other hand, the levels of viral removal by the actual model of CWF would not be effective in decreasing the microbial risk regarding the prevalence and concentration of viruses in surface waters, which has been recently reviewed and showed a prevalence as high as 2-4 logs/L of human adenoviruses (HAdV) in surface waters from different geographical areas worldwide (Bofill-Mas ).
Environmental screening was performed in the Metropolitan Area of Port-au-Prince, the capital of Haiti.Sampling took place in July 2011 during the rainy season.Water samples of 10 L were collected for viral analysis from six sampling points two different sites along the main river (Riviere Grise) that flows through the cities (each in triplicate) that are potentially used at domestic level, one borehole and three different sites (at least in duplicate) at open canals that receive domestic wastewater and discharge directly into the Caribbean Sea.Water samples were analyzed within 6 h after collection to determine Escherichia coli concentration at the University of Quisqueya facilities in Port-au-Prince using direct plate analysis on ChromoCult © Coliform Agar (Merck) (ISO :).On the same day, viruses were concentrated from the 10 L water samples by flocculation using a protocol based on organic flocculation with skimmed milk (Calgua ).The resulting, neutralized floccule was kept at À20 W C until shipment to Barcelona for analysis of viruses by polymerase chain reaction (PCR).Floccules of a total 10 L sample were then centrifuged and resuspended in 10 mL of phosphate buffer before nucleic acid extraction of 140 μL.Viral nucleic acids from concentrates were extracted using the QIAamp Viral RNA Mini Kit ® (Qiagen, Valencia, Spain) following the manufacturer's instructions.Each sample was eluted in a final volume of 80 μL.HAdV, JC polyomavirus (JCPyV) and NoV were quantified using quantitative PCR (qPCR) and quantitative reverse transcription PCR as previously described ( Jothikumar ; Bofill-Mas ; Pal ), and direct samples and their 10-fold dilutions were tested.Standard controls were used in all assays, including negative and positive controls (viral suspensions of human adenovirus 2 (HAdV2) and clinical samples that were positive for NoV), with external controls for inhibition using standard DNA (Albinana-Gimenez ).Positive samples of HAdV by qPCR were amplified by nested PCR (Allard ), and purified using the QIAquick PCR purification kit (Qiagen, Valencia, USA).Both strands of the purified DNA amplicons were sequenced with the ABI PRISM™ Dye Terminator Cycle Sequencing Ready Reaction kit with Ampli Taq ® DNA polymerase FS (Perkin-Elmer, Applied Biosystems, Foster, CA, USA) following the manufacturer's instructions.The results were analyzed using the ABI PRISM 3730 XL automated sequencer (Perkin-Elmer, Applied Biosystems), and the sequences were compared with the nucleotide sequences present in GenBank using the BLAST program of NCBI (National Center for Biotechnology Information; www.ncbi.nlm.nih.gov/BLAST).The production of CWF took place in the ceramic workshop of Josep Matés (Fonteta, Girona, Spain).Several batches of small-scale-sized filters (1:2.5) were produced by following the production instructions from The Ceramics Manufacturing Working Group, and different compositions were tested.After analyzing the microbiological removal efficiencies, a real-scale-size filter was created to test the selected prototype.Small-scale-sized prototypes with different amended oxides and different firing atmospheres (oxidative vs. reductive) were obtained with clay and rice husks from La Bisbal (Girona, Spain).The natural oxides that were selected for testing included Fe 2 O 3 , Fe 3 O 4 and Al 2 O 3 , and these were added at 16% weight into the milled clay before forming the filter.Each clay batch was mixed with pulverized rice husk that was screened with a sieve (250 μm) at 23% weight.Filters were pressed in a plaster mold, removed, stamped with a number and dried before firing at 950-990 W C for 8-9 h in an oxidative atmosphere, except for those produced to test the effect of different firing atmospheres.Those filters with no added oxides and the same proportion of rice husks were fired in oxidative or reductive atmospheres (two replicates per atmosphere) with three different origin clays from La Bisbal (non-calcareous) and two samples from Haiti Artibonite clay (calcic based) and Aquin clay (non-calcareous).
, DNA virus, and the coliphage MS2 as a representative of smaller sized ssRNA viruses (27 nm).Water from the tap (with turbidity <5 NTU and pH 8) was conditioned by adding sodium thiosulfate at 10% to inactivate free chlorine and stocks of challenge microorganisms were added to reach a minimum final concentration of 10 5 IU/mL.To quantify the efficiencies of bacterial reduction, E. coli (Spanish Collection of Type Cultures, CECT 515) was cultured in tryptone soya broth and quantified after filtration and phosphate buffer solution dilution by plating on trypticase soy agar.The viruses spiked were HAdV2 and MS2 as DNA and RNA viral models, respectively.HAdV2 was cultured in A549 cell culture and quantified by qPCR (Bofill-Mas ).MS2 was obtained from the American Type Culture Collection (ATCC 15597-B1), was multiplied by adding MS2 suspension to a log-phase host culture (Salmonella WG49, CECT 4625) and was enumerated following the double agar layer procedure.At the same time, NoV GII obtained from clinical samples was used to spike-challenge the water samples to add evidence for the viral removal of the selected filters.
that were amended with metal oxides and filters fired in different atmospheres (reductive vs. oxidative) were tested in a set of microbiological-removal efficiency studies.Each prototype was tested at least in two parallel filters.After production, the ceramic filters were saturated in mineral water overnight, and blank tests were performed by pouring mineral water through the filter.One liter of the challenge water was filtered in each test at least twice for each prototype.Log reduction values (LRVs) were calculated by subtracting the log 10 of the final concentration from the log 10 of the initial concentration, and mean values and standard deviations were calculated.The Wilcoxon rank-sum test was applied to detect significant differences in LRV between tested filters prototypes vs. the control filters (no added oxides þ oxidative firing); p-values are shown in Table6.The black filter (fired in a reductive atmosphere) with an extra 5-cm layer of ceramic gravel (2-5 mm) was the selected prototype based on the microbiological reduction values of the previous experiments.To evaluate its long-term performance, the WHO-testing recommendations for HWTS (WHO ) were used with two 1:1-scale prototype replicates of fired pieces with a final diameter of 320 mm and height of 230 mm.The performance of the selected models was monitored for over 1,000 filtered liters and a daily dosing of 40 L of dechlorinated, municipal tap water.Challenge waters were spiked with the selected micro-organisms (as in previous experiments) once per week and allowed to filter for 4 h before the effluent was collected to evaluate filter performance.Water quality parameters such as pH and flow rate were measured daily.
results obtained in the water samples in Haiti showed high-levels of human fecal contamination in the superficial waters of the Metropolitan Area of Port-au-Prince during July 2011 and are summarized in Table 2. HAdV were detected in high amounts in water from open canals that were considered to contain urban sewage (4.35 × 10 4 GC/ 100 mL, 6/6), as well as at the two river-water-sampling sites (1.03 × 10 2 GC/100 mL, 4/6) and at the borehole level (3.9 × 10 4 GC/100 mL).These results are in accordance with previous studies, which showed the abundance of HAdV and its use as indicators of human contamination (Bofill-Mas ).Important viral enteric pathogens that produce gastroenteritis were also detected: HAdV types 40-41, NoV GI and NoV GII, which emphasizes the importance of implementing water and sanitation programs that are focused on the reduction of viral infections in the population.

Fe 2 O 3 ,
Fe 3 O 4 and Al 2 O 3 oxides did not significantly increase the efficiency of disinfection when compared with the control filters.The LRVs that were observed were lower than 1 logarithm for all types and for each microorganism that was tested, except for the filters with Al 2 O 3 , which showed a slightly significant increase in E. coli removal of 1.34 LRV (p value 0.046).In contrast, filters that were fired in a reductive atmosphere and produced from three clays of different origins (Spain and Haiti) had greater removal efficiencies for viruses (HAdV, MS2 and NoV) and bacteria (E.coli) around 3.0 log and 2.5 log, respectively, being significant compared to control filters

Figure 1 |
Figure 1 | Pore-size distribution of ceramics that were fired in oxidative (squares) and reductive (rhombus) atmospheres.

Figure 2 |
Figure 2 | Zeta potential distribution of clay from la Bisbal fired in oxidative atmosphere (black line) and clay fired in reductive atmosphere (gray line).
than the current standards and what was previously reported in the literature.Furthermore, they reach the WHO targets for HWTS technologies.Black ceramics, which are clay fired in a reductive atmosphere, seems to increment the internal specific surface area and the Zpotential becoming the most feasible model for achieving advantageous results in viral removal of the proposed models.This model seems to be a low-cost option whose production can be easily adapted for a widespread model High levels of waterborne viruses are present in superficial waters of Port-au-Prince, Haiti, highlighting the importance to improve household water technologies for viral removal efficiencies.A new prototype of CWF fired in a reductive atmosphere, a variation that maintains the product as nonexpensive, improves viral removal efficiencies up to 3 logs, accomplishing the WHO requirements for HWTS technologies.Long-term assays with this new prototype show consistent removal values after 1,000 L filtered.Ceramics fired in a reductive atmosphere show greater surface area and higher Z-potential, the most feasible explanation for viral disinfection properties of this new model.

Table 1 |
Prototypes of the produced CWF

Table 2 |
Microbiological contamination in superficial waters of Port-au-Prince ND: not detected.

Table 3 |
Chemical analysis of clays used to make CWFs

Table 4 |
Mineralogical analysis of ceramics filters fired in different atmospheres in per-

Table 5 |
Specific surface areas of Artibonite and La Bisbal filters that were fired in differ-

Table 6 |
Microbiological effectiveness of different ceramic water filter prototypes in log 10 reduction values and statistical significance levels from Wilcoxon rank-sum test a n ¼ number of assays.b Significance levels of 5% or less (p 0.05).