Laboratory water: choosing the grade, the treatment train and the consumables

What the water grades actually mean, which instrument needs which, why your feed water analysis decides the running cost, and the storage and stagnation problems that ruin the first week back after a long closure.

The blanks will not come down. The ICP baseline creeps upward through the morning. A ghost peak appears in a chromatogram that was clean in March and shifts position depending on the day. In each case the analyst rechecks the standards, then the column, then the operator, and the water is the last thing anyone examines, because it comes out of a tap on the wall and it looks like water.

The grades, and what they actually mean

There are three broad grades, named differently by the standards you are likely to meet in a specification.

• Grade 3, or Type III. Reverse osmosis water. General washing, first rinse of glassware, feeding an autoclave, a water bath or a humidifier.
• Grade 2, or Type II. Reverse osmosis followed by ion exchange or electrodeionisation. Media and buffer preparation, general reagents, and feed to a point of use polisher.
• Grade 1, or Type I. Ultrapure water, polished at the point of use, typically at 18.2 megohm centimetre at 25 degrees Celsius, with low total organic carbon and a final filter.

One distinction matters more than the rest and is the one most often missed: resistivity measures ionic content only. It tells you nothing about organic carbon, nothing about bacteria and nothing about particles. A panel reading 18.2 confirms that the ions have gone. It does not confirm that the water is fit for LC-MS or for cell culture. A system whose only monitor is a resistivity meter is monitoring a third of the problem, so if your work is sensitive to organics, insist on an on-line total organic carbon monitor and budget for it.

On naming: ISO 3696 uses grades 1 to 3, ASTM D1193 uses types I to IV, and clinical laboratories will meet the CLSI reagent water terminology. The schemes are similar but not identical, so quote the standard as well as the grade in any specification you write.

Which instrument needs which grade

Use	Grade	Why
Glassware first wash, cooling baths, autoclave feed	Grade 3	Scale and gross contamination are the only concerns
Media, buffer and general reagent preparation	Grade 2	Ionic purity matters, trace organics usually do not
Final rinse of glassware for trace work	Grade 1	The last thing to touch the glass sets the blank
Atomic absorption and ICP-OES	Grade 1	Trace metal blanks limit what you can report
ICP-MS	Grade 1, with attention to trace metals	The detection limit is set by the blank, not by the instrument
HPLC and UHPLC	Grade 1, low organic carbon	Organics concentrate on the column and reappear as ghost peaks in gradient runs
LC-MS	Grade 1, low organic carbon, freshly produced	Background ions suppress signal and confuse identification
Total organic carbon analysis	Grade 1, dedicated	The blank is effectively the measurement
Cell culture	Grade 1 with ultrafiltration	Endotoxin and bacterial fragments matter even when the water is sterile
Molecular biology	Grade 1, nuclease free	Nucleases survive filtration that removes the organism that made them
Clinical chemistry analysers	As the analyser manufacturer specifies	Analyser performance and warranty both depend on it
Climatic chambers and steam humidifiers	Grade 2 or 3 as specified	Scale is the failure mode, not chemistry

Your feed water is the design input

The most useful thing you can do before asking anybody for a quotation is to have the incoming water analysed: hardness, chloride, iron, silica, turbidity, free chlorine and organic carbon. Without that analysis every supplier is sizing the pretreatment by guesswork, and you pay for the guess in cartridge consumption for the life of the system.

• Municipal supply varies through the year. Hardness and residual chlorine move, and turbidity rises sharply through the monsoon. A system sized on a January sample will struggle in July.
• Deep tubewell supply often carries iron and manganese, which foul membranes quickly and stain everything downstream. Iron removal belongs upstream of the purification system, as building plant, not inside the laboratory unit.
• Almost every building stores water in a roof or underground tank. Whatever grows in that tank is the real feed to your system, so the tank cleaning schedule becomes part of your water quality plan whether anyone intended it or not.
• Free chlorine damages polyamide reverse osmosis membranes. Carbon pretreatment is not optional where the supply is chlorinated, and the carbon has a service life that somebody has to track.

The treatment train, and what each stage is for

1. Sediment filtration, to keep particles out of everything downstream and to stop the carbon fouling early.
2. Activated carbon, to remove free chlorine and some organics before the membrane ever sees them.
3. Softening or antiscalant dosing, where hardness would otherwise scale the membrane.
4. Reverse osmosis, which does the bulk of the work by removing most ions, organics, particles and organisms in a single stage.
5. Electrodeionisation or ion exchange, to remove the remaining ions and hold the water at Grade 2.
6. Storage with recirculation for the Grade 2 water, sized to the peak demand rather than the daily total.
7. Point of use polishing: a mixed bed cartridge, ultraviolet treatment at 254 nm for bacteria and at 185 nm where organic carbon has to be reduced, and a final filter. A 0.22 micrometre filter for particles and bacteria, or an ultrafilter where endotoxins or nucleases matter.

The architectural decision is central against point of use. In a multi room laboratory, a central Grade 2 system feeding several small polishers is usually the better answer, because it puts the expensive, short lived polishing consumables only where Grade 1 is genuinely needed. Producing Grade 1 centrally and piping it around a building means it picks up contamination on the way, and you end up paying polishing costs on water that will wash glassware.

Storage, stagnation and the tank nobody opens

• Do not store Grade 1 water. Make it at the point of use and use it immediately. Ultrapure water is aggressive and begins taking up whatever it touches, including the container you put it in.
• Storage tanks need a vent filter, hydrophobic to keep bacteria out and normally with a carbon dioxide trap, because dissolved carbon dioxide from room air visibly reduces the resistivity of stored water.
• Above a modest volume, recirculation stops being optional. Still water in a plastic tank at Dhaka temperatures grows a biofilm, and a downstream filter does not solve a biofilm, because it sheds continuously into the water passing it.
• Sanitise on a schedule and record it, at whatever interval the manufacturer specifies, plus one extra after any long shutdown.
• Plan for the long closures. A system that stands stagnant through an Eid holiday or a semester break needs a written restart procedure: flush to drain, sanitise if specified, verify the monitors, and do not run the first samples of the term on the first water out of the tap. This is the single most common reason for a bad batch of results in the first week back.

Consumables, lead time and the false economy

• Ask for the annual consumable cost and the lead time for every cartridge before you compare capital prices. A system whose cartridges take months to arrive is a tap with a display on it.
• Confirm the consumables can be imported and cleared without difficulty. Some contain resins or biocides that complicate shipping, and finding that out at the port is not the moment.
• A distillation still is cheap to buy and expensive to run. In hard water it scales, it consumes a great deal of electricity, it produces heat you then pay to remove, and it does not deliver Grade 1 water. Choosing a still because the capital line looks smaller is the clearest false economy in this category.
• Put the water system's own monitors on the calibration schedule. A drifted resistivity cell is worse than no monitor at all, because it produces a comfortable number that nobody questions.
• Size for peak demand, not average. A laboratory that needs twenty litres within ten minutes every morning needs storage capacity, not a larger membrane.

Back to all Insights