High-Purity Water Filtration

It depends where in the train the filter sits, and the sequence matters more than any single stage. Depth and cartridge prefilters upstream remove particulate, suspended solids and carryover from softeners or carbon beds. Activated carbon removes chlorine and chloramine, which is a protection step for downstream membranes rather than a filtration one. Reverse osmosis rejects dissolved ions, organics and the large majority of microorganisms and endotoxin. Final membrane filtration at 0.2 micron retains bacteria. Ultrafiltration, rated by molecular weight rather than micron, is what removes endotoxin reliably.

Sterilising-grade final filtration in pharmaceutical water is 0.2 micron, sometimes written 0.22 micron, and the rating is defined by performance rather than by pore measurement. A sterilising-grade filter is one validated to ASTM F838 by challenge with Brevundimonas diminuta at 10 to the seventh organisms per square centimetre of filter area with no passage. That is why the rating cannot be inferred from a pore size claim alone. Prefilters upstream are typically 1 to 5 micron and exist to protect the final filter and extend its service life.

Not conventional membrane filters, and this is a common and consequential misunderstanding. Endotoxin is a lipopolysaccharide fragment far smaller than a bacterial cell, and it passes a 0.2 micron sterilising-grade filter without difficulty. A filter that retains the organism does not retain what the organism sheds, which is precisely why a system can hold microbial counts within limits and still fail the Water for Injection endotoxin limit of 0.25 endotoxin units per millilitre. Removing endotoxin requires ultrafiltration rated by molecular weight cut-off, or distillation.

The water quality requirements come from the pharmacopoeial monographs, USP and the European Pharmacopoeia, rather than from any filtration standard, and they specify the outcome rather than the equipment. USP general chapter 1231 provides the supporting guidance on system design and microbial control, including its caution about relying on filters in distribution. Filter performance itself is characterised against ASTM F838 for bacterial retention. The hardware, housings, connections and pipework, falls under ASME BPE where the system is bioprocessing equipment.

Most often the filter is being asked to compensate for a system problem it cannot fix. A 0.2 micron filter placed at a point of use to control counts becomes a growth site itself if it is not sanitised on a proper schedule, and it masks the underlying cause rather than removing it, which is why USP guidance cautions against relying on distribution filtration. Other recurring causes are prefilters left in service past their pressure differential limit, carbon exhausted so chloramine reaches and damages the membranes, and integrity testing performed only after use.

By a physical measurement that correlates with pore structure rather than by inspection. The bubble point test pressurises a wetted filter until gas breaks through the largest pore, and a low bubble point indicates an oversized pore or a defect. The diffusive flow or forward flow test measures gas diffusion through the wetted membrane below bubble point and is generally preferred for larger filter areas. Pressure hold or decay measures loss over time in a sealed system. Hydrophobic vent filters, which cannot be water-wetted, use the water intrusion test instead.

They do different jobs and sizing one for the other is a frequent error. Prefiltration, typically 1 to 5 micron depth or cartridge elements, removes bulk particulate and protects what is downstream, and it is a consumable expected to load and be changed on differential pressure. Final filtration at 0.2 micron is a sterilising-grade barrier validated for bacterial retention and integrity tested. Undersizing the prefilter shortens final filter life dramatically and raises cost, because a final filter is several times the price and its change is a qualification-relevant event.

Yes, and the trains differ in a way that matters. For Purified Water the requirement is conductivity, total organic carbon and microbial control, so the filtration train is about particulate and bioburden management. For Water for Injection the endotoxin limit of 0.25 endotoxin units per millilitre applies as well, and that is not achievable with membrane filtration alone: it requires either distillation or ultrafiltration rated by molecular weight cut-off. We also handle the tank vent filters, which are hydrophobic and integrity tested by water intrusion rather than bubble point.

Validation runs at two levels and both are needed. The filter itself is validated for bacterial retention against ASTM F838 and, where the process demands it, for extractables and for compatibility with the actual fluid and sanitisation conditions, which the manufacturer normally supports with documentation. The installation is then qualified as part of the water system: correct components installed as specified, integrity testing demonstrated in place, sanitisation reaching the filter housing, and then the three phase sampling programme that qualifies the system as a whole rather than the filter in isolation.

Usually yes, though three engineering constraints decide it. Physical space and access for the housing and for changing elements, which is frequently the binding limit in a congested plant room. Available pressure, since every filter imposes a differential that the existing pump must accommodate at design flow with the filter loaded rather than clean. And sanitisation compatibility, because the housing and elements must withstand whatever the loop uses, hot water at 80 degrees C, clean steam or ozone, and not every element is rated for all three. We survey these before quoting.

Membranes are typically polyethersulfone or polyvinylidene fluoride for aqueous service, with polytetrafluoroethylene used for hydrophobic vent filters. Housings are 316L stainless steel with sanitary connections, internally finished to suit the service and drainable so they do not hold water between uses. Elastomers are usually EPDM or PTFE-encapsulated silicone chosen for compatibility with the sanitisation method. Every wetted component needs documentation supporting its use in pharmaceutical service, and on a compendial system the housing finish and drainability fall under the same ASME BPE expectations as the pipework.

Two things reliably do, and a third matters more than either. Ultrafiltration, rated by molecular weight cut-off rather than micron and typically in the region of ten to thirty kilodaltons, retains endotoxin. Distillation removes it through the phase change, which is why it remains the conservative route for Water for Injection. The third is prevention: because endotoxin is heat-stable and shed by gram-negative organisms, the durable control is eliminating the biofilm producing it rather than removing the product downstream. A system relying on an endotoxin removal step to compensate for biofilm is fragile.

Yes, and the timing is the part worth agreeing in writing. Sterilising-grade filters should be integrity tested after installation and after sanitisation, and again after use, because a post-use test is what supports the batch that has already been filtered. Testing only before use leaves the possibility of a failure during the run undetected. Results are recorded with the test method, the wetting fluid, the measured value, the acceptance criterion and the filter serial number, so a specific element can be traced to specific production.

Yes, but on a qualified loop the paperwork is usually larger than the pipework. Adding filtration is a change subject to assessment under change control, and the assessment has to consider whether the new component alters flow, pressure, sanitisation effectiveness or drainability anywhere in the system. Depending on the outcome, partial requalification and a fresh sampling round may be required before the loop returns to compendial service. Planning the change against an existing approved protocol, rather than writing one under time pressure afterwards, is what keeps the outage short.

No, and the distinction matters when comparing quotations. Filtration is one stage within a water system. A complete system covers pretreatment, generation by reverse osmosis, electrodeionisation, distillation or membrane WFI production, storage, the distribution loop with its points of use, sanitisation, instrumentation and control, and the three phase qualification programme that proves the whole thing works. A filtration scope delivers and qualifies the filtration equipment. Where a quotation appears low, it is worth confirming which of those elements it actually includes, particularly qualification.