Double-Pass RO and EDI: How Pharma and Boiler Purified Water Is Actually Made

A single RO takes borewell water at 1,500 mg/L TDS and delivers permeate at 20–40 mg/L. That is a remarkable thing for one membrane to do, and for most process uses it is enough. But a pharmaceutical purified water loop needs conductivity below 1.3 µS/cm, and a high-pressure boiler needs silica in the low parts per billion, and no single RO on earth gets there. For those duties the industry has settled on a standard train — RO, then a second RO, then electrodeionisation — and the reasons it is built that way are worth understanding before you specify one.

Why one pass is not enough

An RO membrane rejects 99%+ of salt, but the last fraction of a percent is stubborn and specific. Three things get through:

• Dissolved CO₂. Carbon dioxide is a gas, not an ion, and passes straight through the membrane. On the permeate side it re-forms carbonic acid, giving low pH and conductivity that no amount of membrane will remove. On a typical Indian groundwater with 200–400 mg/L alkalinity, CO₂ passage alone can put 5–15 µS/cm into the permeate.
• Silica. Rejected at 97–99%, which sounds good until you need 10 ppb in the boiler feed and the raw water has 40 mg/L.
• Boron and other weakly ionised species. Un-ionised at neutral pH and therefore poorly rejected.

The fix for all three is the same: raise the pH between the passes.

The second pass, and why the pH goes up

A double-pass RO feeds the permeate of the first pass into a second, smaller RO. Because the feed to pass two is already very clean (low TDS, SDI near zero), the second pass runs at high recovery — typically 85–90% — and its reject is clean enough to return to the first-pass feed rather than to drain, so the overall water loss is small. On its own, the second pass would take 30 mg/L down to perhaps 2–3 mg/L. But the real trick is dosing caustic soda into the inter-pass permeate to take the pH to 8.5–9.5. At that pH, CO₂ converts to bicarbonate — an ion the membrane rejects — and silica and boron ionise into forms that are rejected far better. A high-pH second pass routinely delivers permeate below 1–2 µS/cm and silica below 20–50 ppb. This is exactly the chemistry that makes high-pH RO the answer to silica, applied to polishing rather than to recovery.

The alternative — a degasser or membrane contactor to strip CO₂ physically between passes, or before EDI — is used where caustic dosing is unwelcome, but the high-pH second pass is the more common design in Indian pharma and power.

EDI: the final polish

Electrodeionisation is a mixed-bed ion exchange that regenerates itself electrically. Water flows through a stack of dilute chambers filled with mixed resin, bounded by cation- and anion-selective membranes; a DC voltage across the stack pulls ions out of the resin into concentrate chambers, and at the same time splits water into H⁺ and OH⁻ that continuously regenerate the resin. No acid, no caustic, no regeneration downtime, no waste chemical — just electricity and a small concentrate stream that is usually recycled to the RO feed.

What EDI delivers is resistivity of 10–18 MΩ·cm (conductivity of 0.06–0.1 µS/cm) with silica below 10 ppb, from a feed that is already very good. That last condition is the point: EDI is a polisher, not a treatment step. Its feed must be RO permeate with conductivity under about 20–40 µS/cm, hardness under 1 mg/L as CaCO₃, silica under 0.5–1 mg/L, CO₂ under 5–10 mg/L, no chlorine, no organics above 0.5 mg/L TOC. Feed it worse water and the stack scales, the resin fouls and the modules — which are expensive — fail early. That is why the second pass sits in front of it: not because EDI cannot handle single-pass permeate on a clean feed, but because on real Indian groundwater the single-pass permeate is at the edge of EDI's tolerances for CO₂ and silica, and the second pass moves the plant comfortably inside them.

Where each duty needs to stop

• Process water, cooling make-up, low-pressure boilers: single-pass RO, perhaps with softening upstream. Permeate at 20–50 µS/cm is fine.
• Medium-pressure boilers (40–60 bar): double-pass RO or RO + mixed-bed. The silica limit in the boiler water decides which.
• High-pressure boilers and turbines: RO + RO + EDI (or RO + mixed bed) for silica in the low ppb range. See boiler feed water and DM water for the boiler-side reasoning.
• Pharma purified water (IP/USP): RO + EDI is the modern standard, with a hot-water-sanitisable loop; double-pass RO + EDI where feed CO₂ or the microbial requirement demands it. RO in pharma water systems covers the recovery side.
• Water for injection: Multi-column distillation or, increasingly, RO + EDI + ultrafiltration with hot sanitisation, under the current pharmacopoeial guidance.

What decides the cost

The membrane train is rarely the expensive part; the pre-treatment and the water chemistry are. Softening ahead of the first pass (so the high-pH second pass does not scale with hardness), reliable dechlorination (EDI resin and polyamide membranes both die on chlorine), and TOC control decide whether the plant runs for years or for months. And the whole train's recovery is set by the first pass, which on a high-silica or high-hardness borewell may be the limiting factor for water cost. Every design conversation therefore starts with a complete feed analysis — including silica, CO₂ (or pH and alkalinity), boron and TOC — and works forward from there.

If you are specifying a purified-water or DM-water plant and want to know whether you need the second pass, the EDI, or both, that is a short conversation to have with us — [email protected] or +91-98100 00233.