RO and DI are different mechanisms and are usually combined rather than chosen between. Reverse osmosis removes the bulk of dissolved solids, organics and microorganisms through a membrane; deionisation polishes the remainder to very low conductivity by ion exchange. Most high-purity systems run RO first and polish with DI or EDI. Paul Industries designs, installs and validates the full train — pretreatment through distribution — nationwide.
How reverse osmosis (RO) works
Reverse osmosis pushes feed water against a semipermeable membrane under pressure. The membrane allows water molecules to pass while rejecting the majority of dissolved salts, larger organic molecules, and suspended matter. Because rejection is size- and charge-based rather than chemical, a single RO pass typically removes about 95–99% of total dissolved solids (TDS) along with most microorganisms and endotoxin-bearing material.
RO is a bulk-reduction step. It handles the heavy lifting economically, dropping feed-water conductivity dramatically, but it does not reach the extremely low ionic levels that ion exchange can. RO also produces a reject (concentrate) stream, so recovery and pretreatment — softening, carbon, and antiscalant dosing — matter for membrane life and consistent performance.
How deionization (DI) works
Deionization uses ion-exchange resins to remove dissolved ions. Cation resin swaps positively charged ions (such as calcium, magnesium, and sodium) for hydrogen, and anion resin swaps negatively charged ions (such as chloride, sulfate, and bicarbonate) for hydroxide; the released hydrogen and hydroxide combine to form water. A mixed-bed DI unit blends both resins to drive conductivity very low and resistivity very high.
DI excels at ionic purity but has clear limits. It does not remove non-ionic dissolved organics, particulate matter, or bacteria — in fact, exhausted resin beds can shed organics and host microbial growth. Resins also exhaust and require regeneration or replacement. That is why DI is best used as a polishing step on already-clean water rather than as a standalone treatment for heavily loaded feed.
RO vs DI compared
| Reverse osmosis (RO) | Deionization (DI) | |
|---|---|---|
| Mechanism | Pressure-driven semipermeable membrane | Ion-exchange resin beds |
| What it removes | ~95–99% of dissolved solids, most organics, particles, microbes | Dissolved ions only (very effective) |
| What it misses | A small residual fraction of ions | Organics, particles, and bacteria |
| Typical purity | Broad, high-quality reduction of TDS | Very low conductivity / high resistivity (approaching ~18 MΩ·cm with polishing) |
| Limitations | Reject stream; needs pretreatment; can’t hit ultralow ionic levels alone | Resins exhaust; no barrier to organics or microbes |
| Typical use | Bulk pretreatment / primary purification | Final ionic polishing after RO |
How RO and DI combine in high-purity water systems
In regulated pharmaceutical and biotech facilities, RO and DI are almost never an either/or choice. A common architecture runs pretreatment, then RO to remove the bulk of dissolved solids, organics, and microbial load, followed by electrodeionization (EDI) or mixed-bed DI to polish the remaining ions down to the required conductivity. EDI is often preferred because it regenerates continuously using electrical current instead of chemicals, reducing handling and downtime.
This staged approach lets each technology do what it does best: RO protects and reduces the load reaching the resins, while DI/EDI achieves the final low-conductivity target. Downstream storage and distribution loops — with sanitary design, recirculation, and appropriate sanitization — then maintain that quality up to the point of use so the water still meets specification where it is actually drawn.
Standards & references
- USP <1231> Water for Pharmaceutical Purposes
- The United States Pharmacopeia general chapter that describes water types, quality attributes (including conductivity and total organic carbon), and system design, control, and monitoring expectations for pharmaceutical water. It frames why both ionic and organic purity must be controlled.
- ASTM D1193 Standard Specification for Reagent Water
- Defines laboratory reagent-water grades (Types I–IV) by measurable limits such as resistivity/conductivity, total organic carbon, and other attributes. It provides a common reference for classifying how pure treated water actually is.
Frequently asked questions
Is RO or DI water purer?
When should I use RO instead of DI?
When should I use DI instead of RO?
Does RO or DI remove organics?
Which lasts longer, RO membranes or DI resin?
Can you combine RO and DI in one system?
Does RO or DI remove bacteria and endotoxin?
What is EDI and how does it compare to DI?
Which is cheaper to operate, RO or DI?
Does RO waste water?
Is RO or DI water suitable for pharmaceutical use?
What contaminants does RO remove that DI does not?
Why does DI resin exhaust so fast on some water?
How do you decide between RO, DI, and EDI for a project?
Can Paul Industries build combined RO and DI systems nationwide?
What is the difference between RO and DI water?
Which is purer, RO or DI water?
Can you combine RO and DI?
Does RO remove bacteria?
Which is used for pharmaceutical water?
Need a high-purity water system built or serviced?
Paul Industries designs, installs, and validates process-equipment and sanitary-piping systems for manufacturers nationwide.
Request a Project Quote or call 201-450-8280Maintaining an RO/DI system: what actually goes wrong
Paul Industries services and maintains high-purity RO and deionisation systems for pharmaceutical, biotech and food manufacturers across the United States. The practices below are the ones that decide whether a system holds its specification between services.
The number on the panel does not measure what most people think
This is the single most important thing to understand about DI maintenance. Resistivity and conductivity measure ions. They are completely blind to bacteria, endotoxin and most organics. A deionisation system can read 18 megohm-centimetre – textbook ultrapure by the ionic measure – while being microbiologically unacceptable.
Worse, a DI resin bed is close to an ideal microbial habitat. It offers enormous surface area, ambient temperature, low and intermittent flow, and organic carbon to feed on. Bacteria colonise the bed, form biofilm on the beads, and shed downstream. The resistivity reading does not move, because bacteria are not ions. Systems therefore fail microbial specifications while every gauge on the panel looks correct – and the maintenance programme is what prevents it, because the instrumentation will not warn you.
Failure modes and what prevents them
| Failure mode | What causes it | Prevention |
|---|---|---|
| Microbial colonisation of the resin bed | Ambient temperature, high surface area, stagnation during low demand | Scheduled sanitisation, recirculation to avoid stagnation, microbial trending rather than conductivity alone |
| Chlorine damage to RO membranes | Feed chlorine reaching the membrane | Activated carbon or bisulfite dosing upstream, with breakthrough monitored – carbon beds exhaust silently |
| Resin fouling by organics | Natural organic matter in the feed loading the anion resin first | Pretreatment, and recognising that anion resin usually exhausts before cation |
| Physical bead damage | Pressure surges and rapid valve operation cracking beads | Steady flow, controlled valve sequencing, soft-start pumps |
| Channelling | Bed disturbance or uneven flow creating a preferential path | Correct backwash technique and distributor condition |
| Silica breakthrough | Silica is weakly held and breaks through before conductivity shifts noticeably | Monitor silica separately where it matters – conductivity is a late indicator for it |
| Premature exhaustion after service | Incorrect valve sequencing during regeneration or bottle change | Documented procedures and trained operators – human error is a leading cause |
What to monitor, and what each measurement misses
| Measurement | Detects | Blind to |
|---|---|---|
| Resistivity / conductivity | Ionic breakthrough, resin exhaustion | Bacteria, endotoxin, most organics, and early silica breakthrough |
| Total organic carbon | Organic load, and one of the two compendial water attributes | Microbial identity and endotoxin |
| Microbial count | Bioburden – the failure conductivity cannot see | Results lag by days, so trending matters more than any single result |
| Endotoxin | Pyrogen risk where the water feeds parenteral processes | Not applicable to Purified Water, which carries no endotoxin limit |
| Differential pressure | Fouling, bead damage, channelling | Water quality itself |
| Silica | Early anion exhaustion | Everything else |
The practical rule: conductivity tells you when the resin is spent, not whether the water is fit. Both matter, and only one of them is on the panel.
Sanitisation options
| Method | How it works | Trade-off |
|---|---|---|
| Hot water | Circulated above the growth range | Effective and residue-free, but many resins do not tolerate it – check the resin specification first |
| Ozone | Dosed into storage, destroyed by UV before use points | Strong control for ambient loops; requires UV destruct and verification of absence at the point of use |
| Chemical | Peracetic acid or equivalent, per resin compatibility | Requires a validated rinse-out to a measured endpoint |
| UV at 254 nm | Germicidal, inline | Kills in the beam only – it does not clean an already-colonised bed downstream |
| UV at 185 nm | Reduces TOC by photo-oxidation | Produces ionic species, so it belongs upstream of a polishing bed, not after it |
The maintenance record is the deliverable
Trending is what makes any of this useful. A log that records resin age and change dates, conductivity and TOC trends, microbial results, differential pressure, sanitisation dates and the feed water analysis lets you see a system drifting weeks before it fails a specification. Without it, every excursion is investigated from scratch. This is also what 21 CFR 211.67(c) expects – records of maintenance, cleaning and inspection – and what our preventive maintenance programme hands over.
Service intervals and the two measurements people take in the wrong place
Two practical points that decide whether a maintenance programme actually catches problems.
Test at the outlet, not the source. Feed-water readings tell you what the system is being asked to do; outlet readings tell you whether it is still doing it. Logging outlet quality over time is what reveals gradual resin decline before it becomes an excursion – a single reading tells you almost nothing, a trend tells you when to schedule.
Continuous recirculation beats intermittent circulation. Stagnant water in a DI loop is where microbial colonisation begins, so a loop that flows continuously holds quality better than one that runs on demand. This is the same argument as the microbial section above, from the other direction: the enemy is stillness.
| Component | Typical service interval | What drives it |
|---|---|---|
| Sediment pre-filters | Commonly every 1 to 3 months | Feed water turbidity – a fouled pre-filter shifts load onto everything downstream |
| Carbon filters | Longer than sediment, but monitored for breakthrough | Carbon exhausts silently. Chlorine breakthrough destroys RO membranes with no warning on any gauge |
| RO membranes | Years, driven by differential pressure and rejection trend | Feed quality, chlorine exposure, cleaning frequency |
| Ion exchange resin | Regeneration or exchange on capacity; replacement measured in years | Feed ionic load, organic fouling, physical bead damage |
| UV lamps | Rated hours, not visual condition | A lamp that still lights may be well below germicidal output |
| Sanitisation | On a defined schedule, not on symptoms | By the time microbial results move, the bed is already colonised |
The carbon-filter line deserves emphasis. An exhausted carbon bed gives no external sign – flow is normal, pressure is normal, and conductivity is unaffected because chlorine is not the ion being measured. The first indication is usually membrane damage that has already happened. Monitor for chlorine breakthrough rather than trusting an interval.
RO vs DI water: which does your process need?
Framing this as a choice usually signals the real question is “what quality does my process require, and what train delivers it?” RO alone rarely reaches compendial conductivity; DI alone is quickly exhausted by a feed that has not been through RO, making it expensive to operate and a bioburden risk when resin beds sit idle. The economic answer for nearly all pharmaceutical and biotech duty is RO followed by EDI, which regenerates continuously and removes the resin-regeneration burden entirely.
Should I use DI or EDI after reverse osmosis?
For nearly all pharmaceutical and biotech duty, EDI. Conventional DI resin beds require chemical regeneration, carry an operating burden, and become a bioburden risk when they sit idle between uses. Electrodeionisation regenerates continuously using an applied electrical field, removing the regeneration chemistry and the associated handling and disposal entirely, which is why RO followed by EDI is the standard train for compendial water generation.