Ceramic and carbon filters

Summary

A ceramic pot filter is a vessel of porous fired clay of about 10 l that hangs inside a receptacle with a lid and tap. Water passes through the wall by gravity (1-2 l/h when the filter is new) and leaves behind dirt, almost all bacteria, protozoa and worm eggs. If the ceramic is impregnated with silver, more bacteria die and less biofilm grows. A family thus obtains 10-25 l/day of filtered water, depending on turbidity, without spending firewood to boil it, with a filter that lasts 1-3 years. It does not retain viruses well (typical reduction of 90 %, sometimes less) nor does it remove dissolved chemical substances (arsenic, nitrates, fluoride, salts). If the source water may have fecal contamination (river, pond, open well) or there are outbreaks, disinfect after the filter. A granular carbon stage improves taste and odor and retains chlorine and part of the organic matter, but it does not disinfect. This article also covers ceramic candle filters.

Prerequisites

• Knowledge: knowing how to work clay and fire it in a ceramic kiln (topic “Ceramic kilns”, not yet an article). To disinfect after the filter: Boiling water, Chemical water disinfection with chlorine and iodine or Solar disinfection (SODIS). Preparing granular carbon (crushing, sieving, washing, activation) is covered in Slow sand filter and carbon biofilter.
• Materials (RDIC recipe, ≈ 6 filters per batch):
  • 30 kg of dry, powdered clay. Brick clay or red earthenware (terracotta) pottery clay both work, as long as they pass the tests in step 1. With 8.9-10 kg of burnout material and 12.5 l of water this adds up to ≈ 52 kg of paste, which yields ≈ 6 blocks of 8.0-8.2 kg (RDIC, p. 33).
  • 8.9-10 kg of ground and sieved burnout material: rice husk, untreated sawdust, coffee husk. Particle size between 0.5 and 1 mm.
  • 12.5 l of clean water.
  • For silver (optional, recommended): 100 g of AgNO₃ (≥ 99 %) for 1.5 l of concentrate, enough for about 900 filters. For every 60 filters: 100 ml of that concentrate and 18 l of distilled or rainwater. Do not dissolve all the nitrate in the working water (you would apply 13-14 times the dose). Alternative: commercial colloidal silver (step 17).
  • For each finished filter: a receptacle of 20-40 l with a lid (food-grade plastic, unglazed clay or clay with a lead-free glaze, or wood sealed with beeswax or left unsealed; not paint or varnish), a tap, a support ring for the rim of the pot and a soft-bristle brush.
  • Kiln fuel: RDIC burns about 1.5 m³ of dry firewood per firing of 96 filters.
• Tools:
  • 0.5 mm and 1 mm sieves (wire mesh, N2; tightly woven cloth or a fine basket, N1, with less precision).
  • Tamper or mallet to crush the clay (N0-N1). Scale or matching buckets to measure by volume (N1).
  • Two-piece mold (male and female) made of wood, fired clay or metal, with a lever or jack press. Without a mold, a potter's wheel (N1).
  • Ceramic kiln with a chimney draft that reaches 850-900 °C (N1). Pyrometric cones or a thermocouple if available (N3-N4).
  • 1 l graduated container and a clock or hourglass timed to 1 h (N1).
  • Gloves, goggles and a dust/silver mask (N1-N4 depending on type).
• Time and people: about 3-4 weeks from mixing to a finished filter: 1 day of mixing and pressing, 1-3 weeks of drying, 1-2 days of firing and cooling, 1 day of soaking and flow-rate testing, 3-8 days of drying and silver painting. A workshop of 3-4 people with a press produces about 50 filters a day (Potters for Peace). On the wheel, a potter makes far fewer, with a less uniform wall.

How it works

Why water passes through and microbes do not. RDIC gives pore sizes of 0.2-3 µm (p. 9; on p. 8, 0.6-3 µm citing Lantagne). A bacterium measures around 1-3 µm, a protozoan cyst about 4-15 µm and a helminth egg about 45 µm. Other measurements give larger pores: van Halem, an average of 40 µm by bubble point; Oyanedel-Craver and Smith, 0.02-15 µm. In other words, retention is not just sieving: sedimentation, diffusion and adsorption on the pore walls also act, and all of that needs residence time. That is why a filter with a higher flow rate retains worse. A virus measures about 0.02 µm and only a fraction is retained, by adsorption.

Water alone through plain clay would pass too slowly. That is why the clay is mixed with a burnout material (sawdust or husk), which burns away in the kiln and leaves gaps that shorten the path through the dense clay. As long as they do not form continuous channels from side to side, filtering is still done by the clay between them. From this come the variables that matter most:

Flow rate serves as quality control. You cannot see a 1 µm pore, but you can see how many liters cross the wall in an hour. A too-high flow rate reveals cracks, large pores or connected voids, and shortens the residence time and the contact time with the silver. In a Potters Without Borders study (2013), filters that started above 1.7 l/h fell below 99 % coliform reduction, and it recommends not going below 53 % clay in the mix (without specifying whether by mass or by volume).

Silver. Silver ions (Ag⁺) damage bacterial membranes and enzymes. When the filter is painted with silver nitrate, the ions are reduced inside the ceramic to metallic silver particles, which gradually release traces that inactivate bacteria on contact and slow biofilm growth. Studies do not agree on how much it contributes:

• Oyanedel-Craver and Smith (2008): silver improved E. coli retention; the amount applied mattered more than the way it was applied.
• van Halem (2006): filters from Nicaragua without silver removed coliforms from canal water just as well as treated ones; with high E. coli loads, silver did improve the result. Filters without silver retained the MS2 bacteriophage better.
• Brown (2007), cited by RDIC (p. 8): in the field, filters painted with silver performed similarly to those without it.

Practical conclusion: silver is a reinforcement, not the basis of the filter. A well-made filter without silver is still useful, but disinfecting afterward is advisable.

Expected effectiveness. The row in WHO table 7.8 is “Porous ceramic and carbon block filtration” and also groups candles and industrial blocks together (LRV = log reduction value: 1 LRV = 90 %, 2 = 99 %, 4 = 99.99 %):

Locally made pots usually fall below 1 LRV for viruses. In the field, RDIC measured an average E. coli reduction of 95.1 %, up to 99.99 %, and among its users there was 46 % less diarrhea (Brown and Sobsey, 2006, cited by RDIC). In the laboratory, van Halem obtained 4-7 LRV for E. coli and 0.5-3 LRV for MS2 (a virus indicator); Lantagne (2001) measured 1.8-4.9 LRV of E. coli. Viruses are the weak point: do not take their retention for granted.

Carbon. Activated carbon has an enormous internal surface where organic molecules are adsorbed: taste and odor compounds, residual chlorine, part of the pesticides and solvents. It does not kill microbes. It retains organic matter that feeds bacteria, so biofilm grows inside it and the water can come out with more heterotrophic bacteria than went in (WHO, Heterotrophic Plate Counts, 2003). It hardly retains salts, nitrates or fluoride, and ordinary charcoal barely retains arsenic (verificar). In a household filter it improves taste and reduces certain organic chemicals, but it adds no microbiological safety.

Procedure

The main procedure is the pot filter, in seven parts: A) raw materials and recipe, B) shaping, drying and firing, C) quality control, D) silver and assembly, E) use and maintenance, F) ceramic candle filters and G) carbon stage.

A. Raw materials and recipe

1. Check the clay. Wet a handful, make a coil about 1 cm thick and close it into a ring of about 2.5 cm. Also make a bar of 10 × 2 × 1 cm with two marks exactly 10 cm apart. Dry it in the shade for 7 days and fire it with the next firing. Find out whether the clay comes from an area with arsenic in the groundwater: you will need this in step 13. Success criterion: the ring does not crack when closed. The bar comes out whole, with no cracks or warping, and after firing the distance between the marks is still between 8.8 and 9.2 cm (total shrinkage of 8-12 % (verificar)). If it cracks or shrinks more, mix in fine sand or grog (ground fired clay) in 10 % increments until it passes. If it crumbles or does not bind, mix it with a more plastic clay.
2. Dry and grind the clay. Dry it completely in the sun (clods that break dry and do not squash like modeling clay). Break them up with a tamper on clean ground and pass them through a sieve of about 1-2 mm, outdoors, wearing a mask and with your back to the wind (see Safety). Success criterion: loose powder with no lumps bigger than 2 mm, no roots and no stones.
3. Prepare the burnout material. Grind the husk or sawdust if needed and sieve it: use what passes through 1 mm and stays on 0.5 mm. Always use the same source and the same sieving, because larger particles give more flow for the same mass. No treated, painted or particle-board sawdust. Success criterion: visibly uniform particles, with no splinters bigger than 1 mm and no large amount of fine dust.
4. Test mixes (only the first time or when changing material). RDIC's reference recipe is 30 kg of powdered clay, 8.9-10 kg of ground rice husk and 12.5 l of water: 23-25 % burnout material by dry mass. Potters for Peace works by volume, between 60:40 and 50:50 (clay:sawdust); Potters Without Borders advises against going below 53 % clay. Start at 55:45 or 60:40 and treat 50:50 as the limit. Without a scale, measure with the same bucket. Make 3 mixes 5 points apart (for example 55:45, 60:40 and 65:35), make 3-5 filters from each and carry them through step 11 (and step 14 if possible). Success criterion (this article's practical rule): keep the recipe in which at least 4 out of every 5 pieces give 1.0-2.0 l/h in the first hour.
5. Mix. First mix dry for 10 min (with a shovel on a tarp), covering your mouth. Add the water gradually with a watering can and knead for another ≥ 10 min; then throw the paste against the tarp to remove air. Divide into blocks of 8.0-8.2 kg (RDIC's measure for its mold; adjust to yours with about 5 % extra). Success criterion: uniform color, no streaks of sawdust or dry lumps. A 5 cm ball squeezed in the hand keeps its shape, does not crack at the edge and does not stick to the palm.

B. Shaping, drying and firing

6. Shape the pot.

   • With a press and mold (Potters for Peace, RDIC): line both pieces of the mold with thin plastic bags. Place the block and close until clay overflows through the relief holes; remove the excess and open. Reference dimensions of the Potters for Peace filter: about 28-30 cm across the mouth by about 25 cm high, about 10 l capacity and a wall 10-12 mm thick (Potters for Peace; Engineering for Change).
   • On the wheel (N1): throw a pot of the same shape and the same 10-12 mm wall. Check the thickness with a marked needle at 5-6 points on the base and walls: more than 2 mm of difference creates preferential flow zones.
   • In both cases, smooth the rim with a wet brush, fill gaps with clay, smooth the inside with a soft rib and mark the date and batch. After 3-4 h in the shade (or overnight), scrape the smooth inner skin to open up pores. Success criterion: symmetrical shape, firm rim with no cracks, wall of equal thickness at every point measured. A deformed piece goes back into the mass: it can be recycled while unfired, no longer once fired.

7. Dry. Dry in the shade the first day and then on ventilated shelves, covered if it rains. RDIC needs 7-15 days in the dry season and 15-18 in the wet season; Potters for Peace, 1-3 weeks. Judge by the pieces at the bottom, which dry more slowly. Success criterion: uniform light color, not cold to the touch against the cheek, and it sounds dry when tapped.

8. Fire. Load the kiln with clay spacers; the pots can be stacked in towers of 3-5, mouth to mouth and base to base. Close the door with bricks and seal the gaps with clay. Place test pieces where they can be seen through the peephole:

   • If available, small Orton pyrometric cones: RDIC uses 014 as a warning and 012 as the endpoint (≈ 866 °C; pp. 41 and 44).
   • If not, test rings of the same clay, about 3 cm, strung on an iron rod (N2) or a previously fired clay rod, that can be pulled out through the peephole with a hook.

   Follow this curve:

   • Dehydration: low fire, vents open, up to ≈ 100 °C; hold it for 2 h. Without a thermometer: the smoke comes out white and damp, and a cold sheet of metal, a shard or a piece of glass held in the chimney fogs up; keep going until it stops fogging (verificar). Rising faster bursts the pieces from the steam.
   • Rise: add firewood gradually. Reaching the final temperature takes 8-10 h.
   • Final temperature: 850-900 °C. RDIC fires at 866 °C; Potters for Peace, at about 860-900 °C. Without cones, look for light cherry red (≈ 850-900 °C), viewed with the kiln shaded or at night; do not wait for orange, which is already hotter. Then pull out a ring, let it cool for 5 min and break it. If it has a black core or scratches with a fingernail, keep the fire going for 30-60 more minutes and pull out another; repeat.
   • End of firing and cooling (RDIC, p. 44, no soak): once cone 012 melts or the ring passes, remove the fuel, close the dampers and leave the kiln closed for 9 h. Then open it and let it cool for 24 h before unloading.

   Record firewood, hours and color, and keep rings from a firing that produced good filters to compare color, sound and hardness in the next ones. Success criterion: the broken ring has the same color inside as outside, with no black core (unburned carbon), does not scratch with a fingernail, and sounds like the reference rings. The pots ring like a bell when knuckled and do not stain the hand. Not softening in 1 h of water is necessary but not sufficient.

C. Quality control of each filter

9. Inspection and sound test. Look inside each pot with a flashlight and tap it with a knuckle at 4 points on the rim and 2 on the base. Success criterion: a clear, sustained sound at every point and no visible crack. A dull or cracked sound reveals a crack even if unseen: discard the piece. Also do not accept cracks in the rim longer than 1 cm, because they grow with use.

10. Soaking. Fully submerge the pots in clean water for at least 5 h; overnight is normal. This saturates the pores so all of them are measured the same way. Success criterion: no bubbles come out when removed and the surface looks uniformly dark.

11. Flow-rate test (on every filter, without exception). Place the saturated pot on a stand that lets it drip into a clean container. Fill it to the rim, top it off if it has dropped, and start the clock. At 1 h, measure the water collected with the graduated container (or how much the level has dropped, using a T-shaped ruler calibrated liter by liter on a waterproofed pot from the same mold). Acceptance criterion (the same throughout the article):

    • Accept 1.0-2.0 l/h in the first hour.
    • Between 2.0 and 2.5 l/h, only if the batch's bacteriological test (step 14) gives ≥ 99 %. Without a test, reject.
    • Reject above 2.5 l/h (large pores, connected voids or a crack) and below 1.0 l/h (it filters too little and will end up unused).

    Context: RDIC sets an optimum of 1.8-2.5 l/h and accepts 1.5-3.0 (4.6, p. 46); Potters for Peace designs for 1-2 l/h (Oyanedel-Craver and Smith, p. 927); Potters Without Borders saw coliform reduction fall below 99 % above 1.7 l/h; the CMWG gives 1-3 l/h. These factory figures are measured in the first hour with the filter full and saturated, and are not the average flow rate in use, which is lower (van der Laan et al., 2017). If a batch produces more high-flow-rate rejects than usual, adjust the recipe (less burnout material or finer particles); if more low-flow-rate rejects, the opposite.

12. Localized-leak test. During the flow-rate test, watch the outside at 10-15 min. Success criterion: the wall dampens evenly and drips are spread out. A trickle, a darker wet line or a concentrated drip indicates a crack: discard. Break or pierce the base of rejected pieces so no one uses them as a filter.

13. Arsenic and metals from the clay itself. van Halem measured in the water of a Cambodian filter ≈ 200 µg/l of arsenic in the first week and still 17 µg/l at 12 weeks (WHO guideline value: 10 µg/l). A study on filters from Kandal (Cambodia) estimated ≈ 43.5 mg of arsenic released per filter, 80 % in the first 100 l, and found that washing reduces it (“Arsenic leaching from ceramic water filters”, 2018). Therefore:

    1. If the clay comes from a region with arsenic in the groundwater and you cannot test it, do not use that clay. van Halem recommends testing leaching before choosing the clay.
    2. If there is no alternative, discard at least the first 100-200 l filtered by each filter (verificar): for example, keep it full in the workshop for 3-4 days at 2 l/h before drying and painting it.
    3. If testing is available, measure the filtered water after washing.
    4. It is a chronic risk (years of exposure). A few weeks above 10 µg/l weigh far less than drinking water with fecal contamination: do not abandon the filter over this if there is no other safe water.

    Success criterion: transparent filtered water, with no clay taste; where testing is available, arsenic < 10 µg/l.

14. Batch bacteriological test (where resources allow). From each firing, pass river or pond water through 2-3 randomly chosen filters and analyze E. coli or thermotolerant coliforms in 100 ml of the inflow water and of the filtered water. Without a laboratory, the H₂S presence/absence test detects serious failures, but does not give a percentage. Success criterion: reduction ≥ 99 % relative to the inflow water and, better still, absence of E. coli in 100 ml. The analysis method is a topic for the planned article “Basic water quality testing”.

D. Silver and assembly

15. Dry the approved pieces lying on their side for 3-8 days. Success criterion: they have returned to the light color of dry clay.

16. Prepare the silver nitrate solution (RDIC recipe, p. 49). With gloves and goggles, away from children and food.

    1. Concentrate: dissolve the 100 g of AgNO₃ in 500 ml of distilled water and make up to 1500 ml. Store it in a tightly closed opaque plastic bottle: light breaks it down. This is enough for ≈ 900 filters.
    2. Working solution: 100 ml of concentrate + 18 l of distilled or rainwater = 18.1 l, for about 60 filters. Prepare only as much as you will use.

    Check calculation: 100 ml of concentrate contains 6.67 g of AgNO₃, and silver makes up 63.5 % of that mass (108/170): 4.23 g of Ag. In 18.1 l that gives ≈ 234 mg of Ag per liter. Each filter receives 300 ml, i.e. ≈ 70 mg of Ag: ≈ 47 mg inside (200 ml) and ≈ 23 mg outside (100 ml). If you dissolved the 100 g in 20 l instead, each filter would receive ≈ 950 mg. Success criterion: transparent, colorless liquid. If it turns yellow, becomes cloudy or forms dark sediment, it has spoiled (light or water with chlorides): prepare another batch with distilled water.

17. Paint. Fill a cup marked at 300 ml and at 100 ml. Paint the inside with a brush, spreading it evenly over the base and walls, until the level drops to the 100 ml mark; paint the outside with the rest. Do not dip it: silver is wasted and the inside needs a higher dose. Success criterion: the whole volume absorbed, with no puddles or drips, and the pot dry in the shade within a few hours. Darkening with light is normal.

    • With colloidal silver: van Halem (2006) describes 2 ml of 3.2 % colloidal silver in 250 ml of water (≈ 64 mg of Ag per filter); Potters for Peace also uses 1 ml in 250 ml (≈ 32 mg). In the laboratory, a 45 s dip in 800 mg/l has been used (Oyanedel-Craver and Smith, 2008). Commercial concentration varies: read the label and calculate for 30-70 mg of Ag per filter (Potters for Peace 32-64 mg; RDIC 70 mg).
    • Without silver: the filter still mechanically retains most bacteria and protozoa, but with less margin and more biofilm. Always disinfect afterward and clean the receptacle more often (step 22).

18. Assemble the unit.

    1. Tap: on the receptacle (20-40 l; RDIC's holds 38 l and stores ≈ 26 l with the filter in place), 3-5 cm from the base (verificar) so sediment stays below it, with a rubber or leather gasket and a nut on the inside.
    2. Ring: on the mouth of the receptacle, a support ring (plastic, wood or wrapped rope) that spreads the weight (the full filter weighs about 16 kg), closes the gap between the pot and the receptacle, and slopes outward, so any splash on the rim falls outside the filtered water.
    3. Filter and lid: seat the filter and put on the lid.

    Success criterion: with the receptacle filled up to the tap and the pot empty, nothing drips from the tap or the gasket in 30 min. A glass of water poured over the ring does not enter the receptacle.

19. First loads. On first use, ≈ 30 % of the applied silver comes out; RDIC discards the first 33 l (p. 48). Fill the filter 3-4 times and discard that water (≥ 33 l; more if step 13 requires it). Success criterion: the water comes out clear and with no clay or metallic taste.

E. Use and maintenance

20. Daily use.

    1. Turbid water: if you cannot read text through 10 cm of water in a glass, let it settle for 12-24 h and decant, or strain it through cloth. Above about 50 NTU, pre-settling is needed (SSWM); if turbidity is habitual, it should not exceed 25 NTU (Emergency WASH).
    2. Loading: fill the pot 2-3 times a day without touching the inside with your hands.
    3. Location: covered, in the shade, on a stand at least 30 cm high, out of animals' reach.
    4. Serving: only through the tap, with a clean cup; never hands or ladles in the receptacle.
    5. Final disinfection: if the water comes from a river, pond or open well, or there are outbreaks of hepatitis or diarrhea, add chlorine to the filtered water (dose in the chlorine and iodine article) or boil it.

    Success criterion: transparent water, with no odor, and cool; the receptacle is only opened to clean it. Calculation. WHO/WEDC (technical note 9) gives 2.5-3 l/person/day for drinking and 7.5-15 l in total. For 5 people: 15 l/day for drinking and ≥ 37.5 l/day counting cooking and basic hygiene. The 1-2 l/h flow rate is that of a new filter in the first hour; in use it drops as the level falls and as clogging occurs, and settles at ≈ 0.5-1.5 l/h on average (verificar). With canal water, van Halem saw all her filters drop below 0.5 l/h within 12 weeks. Thus:

    • Clear water (≈ 1 l/h on average, pot refilled 2-3 times): 15 l in ≈ 15 h → 1 filter for the drinking water of 5 people.
    • Turbid water (≈ 0.5 l/h): 15 l takes ≈ 30 h → 2 filters or mandatory pre-settling.
    • For ≈ 37.5 l/day: 2-3 filters, or a 20 l container inverted above the pot (an RDIC variant) that keeps it full and makes use of the night.

21. Clean the pot (when flow rate drops noticeably, or every 1-4 weeks with turbid water).

    1. Wash your hands with soap and cover the receptacle meanwhile.
    2. Lift the pot by the rim and rest it by the rim on a clean basin, with the base in the air, without touching the outer base.
    3. Scrub the inside with a soft-bristle brush and the unfiltered water: no soap or bleach (they clog the pores or leave a taste) and no metal scourer or sand (they wear down the wall and the silver).
    4. Rinse with source water, empty it and put it back in place. Do not scrub the outside; if it needs cleaning, use filtered or boiled water and a cloth kept just for that.

    Success criterion: the inside regains the color of the clay and the flow rate rises noticeably. According to van Halem, the effect is temporary and the flow rate falls again with turbid water.

22. Clean the receptacle and the tap (every 1-3 months, or if slime or an off-taste appears).

    1. Take out the pot as in step 21 and empty the receptacle through the tap.
    2. Scrub the inside of the receptacle with chlorinated water: 15 ml (1 tablespoon) of plain 5 % bleach with no additives in 10 l of water gives ≈ 75 mg/l, on the order of the shock disinfection dose in the chlorine and iodine article (≈ 55 mg/l, WHO/WEDC technical note 3). The CDC gives, for surfaces, 1 tablespoon per gallon (≈ 15 ml in 3.8 l, ≈ 200 mg/l with 5 % bleach). Without bleach, use freshly boiled water.
    3. Open and close the tap with that water inside.
    4. Rinse with chlorinated drinking water or boiled water (not just filtered water) and dry upside down.

    Success criterion: no slimy film when running a finger over it and no moldy smell.

23. Service life and signs of failure. Replace the pot at whichever comes first:

    • Time: every 2 years (RDIC); Oyanedel-Craver and Smith give a typical life of 2-3 years. SSWM recommends every 1-2 years and Emergency WASH 6-12 months with poor-quality water.
    • Breakage: visible crack, cracked sound, a chip at the rim, or a flow rate that suddenly rises without having been cleaned.
    • Wear from cleaning: after many brushings the wall thins. If damp spots appear on the outside when filled that stand out before the rest, treat it as a localized leak (step 12).
    • Filtered water that is turbid or has an odor.

    A broken pot is not repaired with clay, cement or glue: it is retired. Success criterion: a knuckle tap gives a clear sound and, right after cleaning, the first-hour flow rate (step 11) is no higher than 2.5 l/h nor higher than what the filter had when new (recorded on its sheet).

F. Ceramic candle filters

24. What they are. Hollow ceramic cylinders with a threaded nozzle, made of fine clay or diatomaceous earth, with finer pores than the pot (on the order of 0.2-1 µm according to manufacturers (verificar)), sometimes with silver or a carbon core. They are mounted in two stacked buckets:

    1. The candle hangs inside the upper bucket and passes through its base with a rubber gasket and nut.
    2. Water enters from outside to inside the candle and drips through the nozzle into the lower bucket, which has a tap.

    Each candle gives ≈ 0.1-1 l/h (SSWM), so 2-4 are fitted. Its weak point is the gasket: if it leaks, unfiltered water gets through. Assembly success criterion: with the upper bucket full and the candles replaced by a blind plug (or with the nozzle blocked from below), nothing drips into the lower bucket in 1 h.

25. Use and cleaning of the candles. The first 2-3 loads are discarded. When the flow rate drops to half:

    1. Unscrew the candle with clean hands, holding it by the nozzle.
    2. Scrub the outer surface with a soft scourer or brush under filtered or boiled water, without wetting the inside of the nozzle, until the original color reappears.
    3. Reassemble it.

    Replace it if it has cracks, has lost thickness (some manufacturers give a minimum diameter gauge) or, if it contains carbon, when a bad taste returns. Making candles locally requires very fine, uniform clay, cast in a plaster mold or pressed, and the same controlled firing: it is harder than the pot and best left to experienced workshops. Success criterion: after cleaning, the flow rate returns to more than 80 % of the initial value and the water comes out transparent.

G. Carbon stage in the household filter

Preparing granular carbon (crushed to 0.5-2 mm, sieved, washed and activated) is covered in Slow sand filter and carbon biofilter. This section only explains how to integrate it.

26. Set up a carbon bucket before the ceramic (recommended option). Carbon goes before so the ceramic can afterward retain the bacteria its biofilm sheds. A bag of carbon put inside the pot does not work: water leaves through the walls and base without passing through it, and it also blocks filtering surface.

    1. Bucket: one of food-grade plastic or clay, with the base pierced with holes of ≈ 3-5 mm, that rests firmly on the mouth of the receptacle or on two crossed slats, and drips into the pot. The set's lid moves to the bucket.
    2. Bed: at the base, a tightly woven cotton cloth or 2-3 cm of washed fine gravel, and on top of it washed granular carbon in a layer ≥ 10 cm thick (verificar), covering the whole base from wall to wall.
    3. Volume: do not load into the bucket more water than fits free in the pot (≤ 10 l with the pot empty), so it does not overflow.
    4. Contact time: time = bed volume ÷ flow rate through the bed; aim for ≥ 5 min (verificar). Example: a bucket 25 cm in diameter with 10 cm of carbon → a bed of ≈ 4.9 l. If 10 l takes 10 min to come out (1 l/min), the time is ≈ 5 min. If it comes out faster, plug holes with wooden pegs or add more carbon.

    Success criterion: water leaves the bucket only through the holes, after passing through the bed (it does not run along the sides of the bucket or overflow the top), no black grains fall into the pot, and the pot does not overflow with a full load.

    • Secondary option, after the ceramic: a carbon cartridge at the tap or a chamber between pot and receptacle. It gives better taste, but the carbon stays on the clean side and breeds bacteria. Use it only with disinfection afterward (boiling, or chlorine added to the cup and not to the receptacle, because the carbon would remove it) or by changing the carbon often.
    • A bag of carbon inside the receptacle or the pot only improves the taste slightly by contact; it does not count as a treatment stage.

27. Maintain the carbon. Wash new carbon until the water comes out clear, so the black dust does not clog the ceramic. Change it when the taste or odor it used to remove returns, or every 2-3 months (verificar). Spent carbon is regenerated by heating it (see the biofilter article) or discarded. Success criterion: water with no chlorine, mold or pond smell.

Verification

From simplest to most demanding:

1. By sight and smell. A glass of filtered water and another of source water, against a white background. The filtered one must be clearly more transparent, with no particles or odor. If it is just as turbid, there is a crack or side leak.
2. Monthly flow rate. With the pot filled to the rim, measure the liters collected in the first hour (step 11) and record them on a sheet attached to the filter. If it drops, cleaning is needed. If it rises above 2.5 l/h or above the flow rate it had when new without having been cleaned, or rises suddenly, a crack is likely: retire the pot.
3. Sound test at every cleaning (step 9).
4. H₂S presence/absence test (strips or bottles with culture medium) on 100 ml of filtered water, every 1-3 months. A color change to black in 24-48 h at 25-35 °C indicates probable fecal contamination.
5. E. coli count (membrane filtration or chromogenic substrate): 0 CFU in 100 ml of filtered water. With inflow water ≥ 1000 CFU/100 ml, also measure the reduction: ≥ 99 % (2 LRV), better ≥ 99.9 %.
6. Chemicals (with a laboratory): arsenic < 10 µg/l in new filters made of clay of doubtful origin. Silver: WHO sets no guideline value, but considers up to 0.1 mg/l tolerable. In Oyanedel-Craver and Smith's tests (2008), the water came out above 0.1 mg/l at the start and below it after 200 min of continuous use.

None of these tests verifies virus retention. That is provided by subsequent disinfection.

Common mistakes

Safety

• False security: the most serious risk. A cracked, poorly fired filter, or one with a gap at the ring, lets contaminated water through with an almost clean appearance. Diarrhea, cholera or typhoid fever can kill, especially young children. Measure the flow rate and do the sound test regularly. With heavily contaminated water, during outbreaks, or for infants, always disinfect after the filter.
• Viruses. A significant portion of viruses get through: hepatitis A and E, rotavirus, norovirus, poliovirus. Hepatitis E is especially dangerous in pregnant women. In areas with these outbreaks, chlorine or boiling after the filter is mandatory.
• Chemicals it does not remove. Neither the ceramic nor the charcoal reliably removes arsenic, nitrates, fluoride, salt or heavy metals. If the source is suspect (areas with known arsenic, mines, fertilized fields), the filter does not make it safe.
• Arsenic from the clay itself. Some clays release arsenic for weeks: do not use clay from arsenic-affected areas if you cannot test it, and if there is no alternative, discard the initial water (step 13).
• Lead from the receptacle. Never use traditional glazed pottery of unknown origin to store drinking water: low-temperature glazes often contain lead and can leach it. Use unglazed clay or glaze known to be lead-free. Wood should be sealed with beeswax or left unsealed, never painted or varnished.
• Clay dust (crystalline silica). Breathing it for months or years causes silicosis, irreversible, and is linked to lung cancer. Crush and mix dry outdoors, with the wind at your back, wearing a mask (ideally FFP2/FFP3 or N95; at minimum several layers of damp cloth, much less effective) and goggles. Dampen the ground, sweep wet, and do not let children nearby.
• Silver nitrate. Corrosive and oxidizing. In the eyes it can cause serious injury and blindness; on the skin it leaves black stains for weeks and burns if concentrated. Ingested it is toxic, and it is very toxic to aquatic life.
  • Prevention: gloves and goggles; do not store it near fuel, wood or paper; opaque, labeled bottle, away from food and children.
  • Eyes: rinse with clean water for 15 min without stopping, with the eyelid open, and seek medical attention.
  • Skin: wash with plenty of water. If swallowed: rinse the mouth, give sips of water, do not induce vomiting, and seek medical attention.
  • Waste: do not dispose of it in rivers or wells. Precipitate it with common salt (silver chloride, insoluble) and let the sediment dry in a closed container.
• Silver in drinking water. With the dose from step 16 and discarding the first loads, silver stays below the level that concerns WHO. A dilution error (dissolving the 100 g directly in the working water) multiplies the dose by 13-14. Taking excess silver for years causes argyria, a permanent bluish-gray skin discoloration. Do not add “extra” silver, and do not drink colloidal silver solutions.
• Kiln. Burns from pieces or ashes: thick gloves, closed footwear and do not unload before 24 h. In enclosed spaces the smoke carries carbon monoxide, which kills without warning: fire the kiln outdoors or under an open roof, with a tall chimney and away from dwellings. A damp piece can burst: do not look through the peephole during the initial rise.
• Loads and crushing. A block weighs 8 kg and a full filter about 16 kg: lift with your legs and with two people if needed. Presses crush fingers: only one person operates it and no one puts their hands in the mold.
• Bleach. Never mix it with vinegar, ammonia or other cleaners: it releases toxic gases. See the chlorine and iodine article.

Variants

• With fewer resources (N0-N1, no draft kiln):
  • Firing in a bonfire or pit. Reliably reaching 850 °C and holding it throughout the piece is not guaranteed (verificar): soft filters come out, with a black core and erratic flow. If there is no alternative, fire in a pit covered with shards and plenty of firewood, apply the ring, sound and flow-rate tests without exception, and always treat the filtered water as unsafe: the filter only clarifies it, and it is boiled or chlorinated afterward.
  • Without silver. See step 17: this is compensated for with subsequent disinfection and more frequent cleaning.
  • Receptacle without plastic. An unglazed clay jar (well fired, washed and covered), or glazed only if the glaze is known to be lead-free. Never use traditional glazed pottery of unknown origin to store drinking water: it can leach lead. Wooden lid and a clay or cane spout with a wooden plug as a tap. Without a tap, a long-handled ladle reserved for this purpose that never touches the ground; it is worse, because hands end up touching the water.
  • Without sieves. Sift with a fine-weave basket and cotton cloth as a second sieve; the flow rate will come out more variable, so make more test pieces.
• With more resources (N2-N4):
  • Hydraulic press and a two-piece metal mold (RDIC, Potters for Peace): identical wall thickness on every piece. Hammer mill and mechanical mixer.
  • Orton cones 012-014 or a thermocouple. With a thermocouple, firing can include a soak: Oyanedel-Craver and Smith fired at 900 °C with a 3 h soak (p. 928).
  • Calibrated T-shaped ruler to measure the flow rate of dozens of filters at once.
  • Silver nitrate made from metallic silver and nitric acid (N3; releases NO₂: only outdoors or under a fume hood; this is a chemistry topic and is not detailed here).
  • An in-house E. coli laboratory to release batches and analyze arsenic.
  • Iron oxides (laterite, goethite) in the mix to try to adsorb viruses: RDIC tested this and abandoned it because it could not confirm the effect.
  • Industrial candles with silver and a carbon core. For viruses, ultrafiltration with pores ≤ 0.02 µm (≈ 20 nm) or with a certified virus reduction rating under WHO's classification for household treatment (N4). Fibers of 0.1-0.2 µm, like ceramic, need disinfection afterward against viruses.
• Scaling up:
  • Family: one pot filter is enough for the drinking water of 5 people with clear water; two with turbid water or if it is also used for cooking (calculation in step 20). Keep a spare pot at home.
  • Village: a workshop of 3-4 people with a press, 1 kiln of ≈ 100 pieces and a stand for flow-rate tests supplies hundreds of families a year. What decides success is consistent quality: a flow-rate test on every piece, batch records, and teaching each family how to use it; RDIC insists that training matters as much as manufacturing. Selling at cost price favored the purchase of replacements in Cambodia.
  • Schools and health centers: a 20 l container inverted above the pot to get ≈ 30 l per load, or banks of 4-6 filters in parallel.
  • City: the household filter does not scale to a network. For a large population, centralized treatment is used (coagulation, rapid or slow sand filtration and chlorination), and the ceramic filter remains as point-of-use protection when the network is not reliable.

Sources

• Hagan, J.M.; Harley, N.; Hughes, R.; Chouhan, A.; Pointing, D.; Sampson, M.; Smith, K.; Soam, V. Resource Development International – Cambodia Ceramic Water Filter Handbook, v1.3. RDIC / Engineers Without Borders Australia, Phnom Penh, 2009-2010. Tamaño de poro (págs. 8-9), mezcla (4.2), cubos (4.3), prensado (4.4), secado y cocción con conos Orton 012/014 (4.5, págs. 41 y 44), prueba de caudal (4.6, pág. 46), plata (4.7, págs. 48-49), embalaje (4.8), seguridad laboral (5.1). http://potterswithoutborders.com/wp-content/uploads/2011/12/RDIC-Ceramic-Filter-Manual.pdf
• Ceramics Manufacturing Working Group. Best Practice Recommendations for Local Manufacturing of Ceramic Pot Filters for Household Water Treatment, 1.ª ed., CDC, 2011. Intervalo de caudal de 1-3 l/h, consultado de forma indirecta.
• Oyanedel-Craver, V.A.; Smith, J.A. Environ. Sci. Technol. 42(3): 927-933, 2008. Caudal de diseño de Potters for Peace (1-2 l/h), poros de 0,02-15 µm, cocción a 900 °C con 3 h de meseta, vida útil de 2-3 años, dosis de plata y plata en el efluente.
• van Halem, D. Tesis de máster, TU Delft, 2006. LRV de E. coli (4-7) y de MS2 (0,5-3), filtros sin plata, plata coloidal (2 ml al 3,2 % en 250 ml), lixiviación de arsénico, poro medio por burbuja y caída de caudal.
• van der Laan, H. et al. «High flow ceramic pot filters». Water Research, 2017. Caudal inicial frente a caudal en uso y retención.
• «Arsenic leaching from ceramic water filters». Environ. Sci.: Water Res. Technol. 4(2), 2018. Arsénico liberado por filtro (≈ 43,5 mg; 80 % en los primeros 100 l) y efecto del lavado.
• Lantagne, D.S. Investigation of the Potters for Peace Colloidal Silver Impregnated Ceramic Filter. Report 1. Alethia Environmental para USAID, 2001. Citado a través de van Halem y RDIC.
• Brown, J. (2007), citado por RDIC (pág. 8): rendimiento en campo de filtros con y sin plata.
• OMS. Guidelines for Drinking-water Quality, 4.ª ed. con 1.ª y 2.ª adendas, 2022, tabla 7.8.
• OMS. Evaluating household water treatment options, 2011.
• OMS. Heterotrophic Plate Counts and Drinking-water Safety, 2003: crecimiento bacteriano en carbón.
• OMS / WEDC. Nota técnica 9, «How much water is needed in emergencies»: 2,5-3 l para beber, 7,5-15 l/día.
• CDC. Limpieza y desinfección con lejía: 1 cucharada por galón para superficies.
• SSWM. Ficha «Colloidal Silver Filter»: caudal de vela (0,1-1 l/h), sustitución cada 1-2 años, turbidez máxima de 50 NTU.
• Emergency WASH Compendium. «H.3 Ceramic Filtration»: turbidez (25/50 NTU) y vida de elementos.
• Potters for Peace, «Ceramic Water Filter Project», y Engineering for Change, «Silver Filter»: medidas de la maceta, cocción a ≈ 860-900 °C, plata coloidal (1-2 ml en 250 ml) y producción de un taller.
• Potters Without Borders. «Study on the flow rate of ceramic filters», 2013: arcilla ≥ 53 % y umbral de 1,7 l/h.