Feed water is the one input a clinic rarely inspects and almost never owns outright. It arrives through a building line, passes through whatever treatment the landlord or the previous tenant installed, and ends up as steam inside a chamber where the consequences of a bad batch show up as stained pouches, failed indicators or a postponed list. The three treatment options most often proposed to dental and medical clinics — softening, reverse osmosis and deionisation — are not alternatives on a quality ladder. They remove different things, fail in different ways, and leave behind completely different maintenance records. This guide compares them on what each one actually changes in the water, and on the maintenance burden that follows, using the Canadian limiting values and the AAMI water-quality framework as the two reference points. If you are still choosing the packaging that goes into the chamber, the Class 4 dual-indicator pouches are where most water complaints become visible first.
What the two guidance sets actually specify
Two documents matter, and they answer different questions.
The Canadian reference is the limiting-values list published in Public Health Ontario's reprocessing guidance, adapted from CAN/CSA-Z314.8-08 and CAN/CSA-Z314.3-09 — editions that have since been consolidated into CSA Z314:23. The list is short and specific: pH 6.5 to 8; evaporation residue no more than 15 mg/L; conductivity no more than 50 µS/cm; hardness no more than 0.1 mmol/L; chloride no more than 3 mg/L; iron no more than 0.2 mg/L; phosphate no more than 0.5 mg/L; silica no more than 2 mg/L; cadmium no more than 0.005 mg/L; lead no more than 0.05 mg/L; other heavy metals no more than 0.1 mg/L (PHO, 3rd edition, May 2013, p. 77).
The second reference is ANSI/AAMI ST108:2023, Water for the Processing of Medical Devices, which defines three water categories rather than one list: critical water (final rinse after high-level disinfection, final rinse for critical devices before sterilization, and feed water for process steam), utility water (flushing, washing and intermediate rinsing), and steam tested as condensate. It also sets routine monitoring at monthly for critical water and quarterly for utility water and steam, with annual validation adding ionic and organic parameters (published summary of ST108:2023 Table 2).
ST108:2023 is not a Canadian legal requirement. It is the framework most often cited by accreditation bodies in the United States, and it is useful here for one reason: it separates the water used for the final rinse from the water used for washing, which the single Canadian list does not. A clinic in Ontario or British Columbia works from the provincial list and its sterilizer manufacturer's specification; a clinic with multi-jurisdiction accreditation may have to satisfy both.
| Parameter | Canadian limiting value (PHO, adapted from CSA Z314.8-08 / Z314.3-09) | ST108:2023 critical water (monthly) | ST108:2023 utility water (quarterly) |
|---|---|---|---|
| pH | 6.5 – 8 | 5.0 – 7.5 | 6.5 – 9.5 |
| Conductivity | ≤ 50 µS/cm | < 10 µS/cm | < 500 µS/cm |
| Hardness | ≤ 0.1 mmol/L | < 1.0 mg/L | < 150 mg/L |
| Total alkalinity | Not listed | < 8 mg/L | < 400 mg/L |
| Chloride | ≤ 3 mg/L | < 1 mg/L (annual validation) | < 250 mg/L (annual validation) |
| Iron | ≤ 0.2 mg/L | < 0.1 mg/L (annual validation) | < 0.1 mg/L (annual validation) |
| Phosphate | ≤ 0.5 mg/L | < 5 mg/L (annual validation) | < 1 mg/L (annual validation) |
| Silica | ≤ 2 mg/L | < 50 mg/L (annual validation) | < 1 mg/L (annual validation) |
| Evaporation residue | ≤ 15 mg/L | Not listed | Not listed |
| Bacteria / endotoxin | Not listed | HPC < 10 CFU/mL; endotoxin < 10 EU/mL | HPC < 500 CFU/mL |
Two things stand out in that table. First, the conductivity gap between the Canadian list (≤ 50 µS/cm) and critical water (< 10 µS/cm) is a factor of five, and it is the single number that decides whether a softener alone can ever be sufficient. Second, the Canadian list is a feed-water list — it does not distinguish washing water from final-rinse water, so the stricter of the two readings usually wins in practice.
What each technology removes, and what it leaves behind
The three technologies are ion-selective, and that selectivity is the whole engineering story.
A water softener is a sodium-form cation exchange resin. It swaps calcium and magnesium for sodium, which stops scale forming in the steam generator — but nothing else changes. A manufacturer's own manual puts the limitation plainly: "the remaining contents are not significantly affected by softening, and softening therefore cannot be used as a general method for increasing quality," and adds a second drawback — "the concentration of alkaline bicarbonate in the water increases more than is desirable," which later forms gases in steam generation and degrades steam quality (Getinge GEE series user manual, archived full text at archive.org).
That is why softened water can pass a hardness test and still fail a conductivity limit. The hardness ions are gone; the anions that carry most of the conductivity — chloride, sulfate, bicarbonate — are still there, and the sodium that replaced the calcium keeps the total dissolved solids roughly where they were.
Reverse osmosis pushes water through a membrane that rejects most dissolved ions, along with organics and, with a properly designed system, bacteria and endotoxin. It also strips alkalinity, which makes the product water aggressive to metal if it is not handled correctly; the same manual warns that oxygen-rich deionised water "is aggressive and will attack the materials of the steam generator and feed water system," and recommends de-aeration.
Deionisation exchanges ions for hydrogen and hydroxyl across resin beds. It reaches the lowest conductivity of the three and is the only one that reliably gets below 10 µS/cm, but it removes nothing biologically, and the resin is poisoned by the same contaminants that foul any exchanger: the manual's inlet limits are iron 0.5 mg/L, manganese 0.02 mg/L and free chlorine 0.05 mg/L.
| What needs removing | Softening (Na cation exchange) | Reverse osmosis | Deionisation (ion exchange) |
|---|---|---|---|
| Calcium and magnesium hardness | Yes — the only thing it is designed to do | Yes | Yes |
| Conductivity / total dissolved solids | Largely unchanged; sodium replaces calcium | Yes, typically the bulk of dissolved ions | Yes, to the lowest levels of the three |
| Chloride, sulfate, nitrate | No | Yes | Yes |
| Alkalinity / bicarbonate | No — and bicarbonate can rise | Yes, alkalinity is stripped | Yes |
| Silica | No | Partly; rejection is lower than for chloride | Yes, until resin exhaustion |
| Iron and manganese | No — can foul the resin | Yes, with pretreatment | Partly; fouls the resin |
| Bacteria and endotoxin | No | Yes, with proper design and loop maintenance | No — resin is not a barrier |
| Dissolved oxygen | No | No | No — de-aeration is a separate step |
| Chlorine | No | Must be removed before the membrane | Must be removed before the resin |
The practical consequence is a sequence, not a choice. Reverse osmosis should be preceded by softening — the manufacturer's instruction is explicit that "reverse osmosis should always be preceded by a water softener, which will increase the life of the osmosis equipment" — and deionisation should be preceded by both, because chlorine and iron destroy resin. A clinic that installs a softener and stops there has solved scale and nothing else. The consumables that record the outcome — 24-hour biological indicators and indicator tape — are where a water problem becomes a documented failure.
The maintenance burden, item by item
Maintenance is where the three options diverge most, and it is the part that rarely appears in a quote.
| System | Recurring task | What triggers it | What you record |
|---|---|---|---|
| Softener | Salt replenishment | Level check on a fixed interval; hardness breakthrough on testing | Salt added, date, hardness reading before and after regeneration |
| Softener | Resin inspection and cleaning | Iron fouling, pressure loss, declining capacity | Resin condition, cleaning method, capacity test result |
| Softener | Regeneration settings review | Change in feed hardness or demand | Setpoints, regeneration frequency, brine draw check |
| Reverse osmosis | Pre-filter and carbon change | Scheduled interval or pressure differential rise | Cartridge part number, change date, differential pressure before and after |
| Reverse osmosis | Membrane performance check | Declining rejection rate or product flow | Feed and product conductivity, rejection percentage, temperature, product flow |
| Reverse osmosis | Membrane cleaning or replacement | Irreversible fouling or rejection below the acceptance floor | Cleaning agent, soak time, post-clean rejection, replacement serial |
| Reverse osmosis | Storage tank and distribution loop sanitation | Fixed interval, plus any positive microbial sample | Sanitisation date, agent and concentration, sample results |
| Reverse osmosis | Chlorine monitoring at the inlet | Continuous or each shift where chlorinated feed is used | Free chlorine reading, date, corrective action |
| Deionisation | Resin replacement or regeneration | Conductivity breakthrough above the acceptance limit | Conductivity at breakthrough, resin batch, replacement date |
| Deionisation | Inlet guard checks | Iron, manganese or chlorine excursion upstream | Reading, date, action taken |
| All three | Feed-water quality test | Daily or per the clinic's written schedule for the parameters you can test in house | pH, conductivity, hardness, chloride, tester and date |
| All three | Annual validation | Once a year, covering the parameters you cannot test in house | Full panel results, laboratory name, acceptance statement |
Two entries in that table are worth pausing on. The reverse-osmosis rejection rate is the single most useful number a clinic can track, because it moves before the water fails: a membrane that has begun to bypass shows a falling rejection percentage while the product water still looks clear. And the deionisation breakthrough conductivity is the record that turns a subjective "the water looks fine" into a dated fact.
How each system fails, and what the failure looks like
Softening fails quietly. Hardness climbs back over weeks as resin capacity declines, and the first visible sign is usually scale on the chamber or a white film on instruments — long after the resin stopped doing its job.
Reverse osmosis fails in stages. A chlorine breach punctures the membrane and rejection drops sharply; scaling from inadequate softening raises differential pressure and cuts product flow; a fouled pre-filter passes sediment that the membrane then has to handle. None of these produce a dramatic alarm, which is why the log matters more than the indicator light.
Deionisation fails at a predictable point. The resin exhausts, conductivity rises, and the rise is fast once it starts. The trap is the upstream one: a chlorine or iron excursion damages the resin before conductivity moves at all, so inlet guard records are the only early signal.
There is also a system-level failure that no single device causes. The 2025 Whitehorse General Hospital incident is the clearest published Canadian example: instruments emerged from sterilizers with residue and staining, 135 cataract surgeries were cancelled between late January and early April, and an independent review attributed the problem to the use of "utility water" rather than "critical water" during the final washing of instruments, compounded by the Yukon's naturally hard water. The hospital's response included installing deionisation and ultrafiltration systems (Hospital News). The lesson is not that any one technology would have prevented it; it is that the final-rinse stage and the feed-water stage were treated as the same water when they are not.
What a clinic can test in house, and what it cannot
Conductivity, hardness and chloride are within reach of a clinic willing to buy meters and reagents and to train one person to run them consistently. Conductivity is the easiest and the most informative: it responds to almost every treatment failure except hardness breakthrough on softened water. Hardness titration catches resin exhaustion. Chloride titration catches the anion load that softening leaves behind.
Evaporation residue, iron, heavy metals, silica, phosphate and the steam purity tests — non-condensable gas, dryness value and superheat — need either a laboratory or a specialist steam survey. Those are the parameters to put on the annual validation rather than on a weekly sheet, and the ones to reach for when a chamber is staining but conductivity looks acceptable.
Does softening alone meet a critical-water specification?
No. Softening removes hardness cations and leaves the anions that carry most of the conductivity, so product water typically stays close to the feed's conductivity while hardness drops. Since critical water is defined at below 10 µS/cm and the Canadian list caps conductivity at 50 µS/cm, a softener can satisfy the hardness column and fail the conductivity column in the same sample.
How much maintenance does each option really add?
Measured in tasks rather than money: a softener adds salt, hardness testing and periodic resin care; reverse osmosis adds pre-filter changes, chlorine monitoring, rejection tracking, membrane cleaning and loop sanitation; deionisation adds conductivity breakthrough monitoring and resin replacement, plus the inlet guard checks that protect the resin. The option with the fewest tasks is softening, and it is also the one that changes the least.
What records should a clinic keep for feed-water treatment?
At minimum: the daily or scheduled feed-water test with the parameters you can measure, the date and result of each resin, filter or membrane change, the rejection rate or conductivity trend that justified the change, sanitisation dates for any storage and distribution loop, and the annual validation report covering the parameters you cannot test in house. Those records are also what an inspector reads first, because they show whether the treatment is being run or merely installed.
Choosing a sequence rather than a product
The defensible configuration for a clinic that needs critical-grade water is a sequence: softening to protect the membrane, reverse osmosis to remove the bulk of dissolved ions and alkalinity, deionisation polishing where the conductivity target is below what the membrane can reach, storage and distribution designed to keep the water microbiologically stable, and a de-aeration or materials decision that accounts for how aggressive low-oxygen water is.
Each stage has its own failure mode, its own record and its own interval. The comparison above is not a ranking — the right answer depends on the feed water, the sterilizer manufacturer's specification and the water category the clinic's final rinse requires. What does not vary is the documentation: a clinic that cannot show a dated conductivity reading, a membrane rejection trend and an annual validation panel has no way to demonstrate that the system it installed is still the system it designed.
For the cycle-side records that sit alongside the water log, the compliance log centre has printable templates; if a failed cycle has already happened, the sequence in our guide to water quality and sterilization failures is the one to follow before changing any equipment.
Ordering for a clinic, a lab or a care home? Wholesale and multi-site ordering covers account setup and case quantities, and the B2B wholesale collection lists the lines stocked for institutional buyers. Reselling? become a distributor.
Take this further: print the free sterilization and water-quality log templates, generate a cycle log with the compliance log tool, or ask a compliance specialist to review your setup.
CliniEco Medical is a licensed medical device establishment (MDEL #35334).
Related Reading
- Steam sterilizer water quality: what CSA and AAMI guidance requires — the framework this comparison sits inside
- Sterilizer water quality: why tap water shortens autoclave life — the feed-water failure modes in a clinic setting
- Autoclave water quality in Canada vs the US — where the two specifications diverge
- Weekly autoclave tasks: seals, trays and water quality — the schedule these records belong to
Frequently Asked Questions
Can a water softener replace reverse osmosis for an autoclave?
No. Softening exchanges calcium and magnesium for sodium and leaves the anions that carry most of the conductivity, so it controls scale but not dissolved solids. Reverse osmosis should also be preceded by softening to protect the membrane, so the two work in sequence rather than as substitutes.
Why does softened water still fail a conductivity limit?
Because ion exchange for softening is charge-equivalent: sodium replaces calcium, and the total dissolved solids stay roughly the same. Conductivity tracks total ionic content, not hardness, so a sample can pass hardness and fail conductivity at the same time.
How often should reverse-osmosis membranes be replaced?
On performance, not on a calendar. Track rejection rate, product flow and differential pressure; replace or clean when rejection falls below the acceptance floor you set with the system supplier, and investigate any sudden drop because it usually means a chlorine breach or a failed seal.
Does deionisation remove bacteria or endotoxin?
No. Deionisation removes ions only and is not a microbiological barrier. Where bacteria or endotoxin matter, they are controlled by the reverse-osmosis stage, by storage and distribution design, and by a sanitisation schedule with sampling.
Is ANSI/AAMI ST108:2023 required in Canada?
No. Canadian clinics work from provincial limiting values and the sterilizer manufacturer's specification; ST108:2023 is a United States framework that separates critical, utility and steam water and is widely referenced by accreditation bodies there.
What water tests can a dental clinic run in house?
Conductivity, hardness and chloride are practical with meters and titration reagents and one trained person. Evaporation residue, iron, heavy metals, silica, phosphate and the steam purity tests need a laboratory or a specialist steam survey.
What is the first sign that a water treatment system has failed?
Usually a trend rather than an event: rising conductivity, falling rejection rate, longer regeneration cycles or increasing differential pressure. Visible staining and residue appear later, which is why the log is the early-warning system.