After the latest update of the EP, coming into effect 2017, that includes membrane production of WFI, there is no further regulatory impediment to widespread production of WFI without thermal distillation in US and Europe.

Why has it taken so long for the EP to harmonize with the USP on this subject?
In 2002 the EMEA stated that membranes were not an option for WFI production. This was due to the potential risks associated with the membrane production method, mainly biological fouling of the membrane^(2,3,4). In 2008, the EMEA published a reflection paper on WFI prepared by reverse osmosis (RO) ⁽⁵⁾. This publication stated that the major problem for production of WFI with RO is the microbiological aspect, and the “net effect is that the RO membrane will become, in practice, a bacterial fermenter”. As these concerns were completely met by the thermal process of boiling, evaporation and condensation, thermal distillation was thought to be the only method to reliably produce WFI. It would seem that hard data was missing.

The pivotal change in thinking took place after a large US survey of pharmaceutical water system users, the results were published in 2011⁽¹⁾. The survey questions were sent to users of water systems that make WFI quality water. The extensive survey, performed by Dr. Anthony Bevilacqua and Dr. Teri C. Soli, was a turning point in the field. The survey asked a number of questions regarding the different performance statistics of the system, including: what are the typical microbial levels and endotoxin levels.

In answer to the typical microbial count question, over 90% of the responses claimed that they were achieving WFI levels of below 10 cfu/100ml. This is especially interesting as less than 15% of the systems were characterized as WFI grade.

In November 2011, a mandate was given the Ph. Eur. Water Working Party to review the production section of the Water for Injections monograph (0169) to consider the inclusion of currently available technologies and evaluate whether additional online monitoring is needed⁽⁶⁾. This mandate resulted in organization of an EDQM expert workshop on “Water for Injections – Potential Use of Membrane Systems for the Production” organized in March 2011. The final decision for allowing WFI production by non-distillation technologies was published in a press release on March 2016⁽⁷⁾. The press release stated that: “Any non-distillation technology for producing WFI should be equivalent in quality to that produced by distillation, where equivalence in quality does not simply mean compliance with a specification but also takes into account the robustness of the production method” (underline by the author).

The robustness of water production with RO membranes is high, as the cut off size of an RO membrane is far below the size of a bacterium. In practice, bacteria will pass from the concentrate to the permeate side. The downstream contamination to the RO is rationalized by an imperfection in the membrane or less than perfect sealing gaskets between feed and product⁽⁸⁾. Even during proper operation, biofilm can be form on the product side of a RO membrane and will result in biological growth in the permeate. For Purified Water (PW) this low number of Colony Forming Units (CFU) will not usually cause Out of Specification (OOS) results. As WFI has a much lower limit on the CFU levels, this biofilm usually will push the results OOS. The EP states “Correct operation monitoring and maintenance of the system are essential”⁽⁹⁾. In view of the above, this statement is clear. To minimize product contamination, it is crucial to keep the microbial levels of the RO feed water to a minimum as high levels of incoming bacteria will result in high levels of biofilm downstream. The engineering challenge is to achieve the minimum levels of bacteria but also to minimize maintenance and down time.

A different method of pretreating RO feed water can meet this ideal.
The system will reduce incoming bacterial levels without need for chemical or thermal sanitizations. There is no media for softening or carbon filter for chlorine removal. In addition to removing the areas of possible contamination build up, the pretreatment actively reduces any bacterial levels incoming from the city water.

1. Background
WFI Specification
The specifications of PW EP and WFI EP both overlap and differ. Typically, the chemical parameters of both PW and WFI are easily achieved. In contrast, the microbial targets of both the PW and WFI are more challenging. The inlet water to pharmaceutical PW/WFI systems must meet potable water standards⁽¹⁰⁾. The bacterial specification for the potable water is defined by the relevant organizations from the EU, US, Japan or the WHO. The bacterial value is typically defined as less than 500 CFU/ml. The bacterial levels, after every component in the system, are not officially defined, but it is safe to assume that the level should be equal or less than the incoming water to the system. The EP PW microbial specification calls for less than 100 CFU/ml which can be achieved reliably with single pass RO as long as the RO feed water bacteria level is controlled. The EP WFI microbial specification calls for less than 10 CFU/100ml which the EP has stated that is not reliably achieved with single pass RO even if the RO feed water bacteria level is controlled.
This is the reason that the EP has specified the production equipment needed to generate WFI and not just the final product specification. The EP has specified a minimum of two membrane barriers for WFI production. Either double pass RO or a single pass RO with additional Ultra Filtration (UF).

Meeting Specifications
Data on double pass RO will be presented showing bacterial levels after the first and second passes. The bacterial levels through the pretreatment stages and the production RO stages have to decrease in order to achieve the low levels of bacteria need to meet the WFI specification reliably. Even if the WFI quality is being achieved by the final production process, increasing levels of bacteria as the water advances through the system demonstrates possible future loss of control and OOS occurrences⁽¹¹⁾. Areas of concern in pretreatment systems are softener resin and carbon filters. Bacteria can clump around resin beads causing channeling and bypassing the flow around the contaminated area. The slow flow areas, outside the channels, will enlarge the bacterial mats even more and will cause a vicious circle of accelerated biofilm bloom. Carbon filters are even more susceptible than softener resin to rapid biofilm formation as the carbon is also a source of nutrition and not just an area of slow flow with a high surface area. When the feed water has high average CFU levels, intensive routine maintenance is needed to control the contamination in the pretreatment. Possible microbial control plans can include regeneration with biocides or full sanitization with hot water. The risk of contamination is also heightened with high feed water temperatures.
Microbial control in softener based systems is limited in efficacy as it depends on the incoming bacterial levels which fluctuate per season. The issue is prime as it has the potential for sanitization down time which can cause production shutdown.

2. Media free pretreatment
Electrolytic Scale Reduction (ESR), Hydro Optic Dechlorination (HOD), Hot Water Sanitizable (HWS) RO-RO-Electro Deionization (EDI)

ESR
The ESR is a Stainless Steel (SS) pipe/reactor that dissociates some of the water molecules into OH^– and H⁺ ions⁽¹²⁾. A central anode passes an electrical current through the water to the cathode. Some of the water molecules disassociate to H⁺ and OH^– ions. The inside of the reactor cylinder, the cathode, will have a high pH as a result of the high concentration of OH^– ions.
As pH is a key factor in hardness precipitation⁽¹³⁾, scale will form on the inside surfaces of the reactor and is effectively removed from the flow. This reduction of scale in the RO feed water will retard ion precipitation in the RO concentrate. The anode will have a high concentration of H⁺ ions around it, lowering the pH. This low pH will activate the background feed water chlorides keeping a low level of free chlorine in the system. As long as the ESR is in operation, residual levels of free chlorine will be generated and circulated, acting as a constant disinfectant and actively reducing bacteria from the inlet water levels.

The chemical reactions are described by the following equations:

The Cl₂ will react with the OH^– and form HOCl.

HOD
High intensity UV radiation is an effective process for removing free chlorine^(14,15). The reaction breaks up the free chlorine in its hypochlorous acid form into components that are harmless to the downstream RO and EDI.
The chemical reaction is described by the following equation:

Modern medium pressure UV units can easily generate dosage levels of 1,500,000-1,800,000 µJ/cm². This dose will break up 1 ppm of free chlorine at the unit inlet, to below online detection levels at the unit outlet. Specification for UV disinfection units are in the range of 30,000 µJ/cm² – 60,000 µJ/cm². In view of the above, the UV dechlorination dose has a disinfection overkill of 30-50 which makes it very effective in disinfection in addition to the desired dechlorination. Like the ESR, the HOD has a constant disinfection effect and actively reduces bacteria from the inlet water levels. Both the ESR and HOD have the bacteria reduction effect without need of down time and without need of sanitization. This is known as Continues Bacterial Reduction (CBR) which is system based inherent destruction of bacteria via continuous routine operation without down time.

HWS RO-EDI
RO and EDI have become the standard for pharma water production processes. HWS RO and HWS EDI have been implemented for critical microbial control⁽¹⁶⁾. The standard implementation of HWS RO and EDI is common but usually these systems are sanitized in isolation to the pretreatment. It is not common to heat sanitize softeners and carbon filters with the RO as this can cause particulates, fines and endotoxin to slough off the media and to clog the delicate membranes. The usual practice is not to sanitize the pretreatment or to sanitize offline without combining the RO and EDI in the process.

ESR-HOD-RO-EDI
The combination of the above water treatment units can be combined into a full, integrated, water production system. The pretreatment ESR and HOD will retard scale precipitation in the RO. The HOD will destroy any free chlorine carry over from the municipal supply and will also breakup the free chlorine added by the ESR scale precipitation process. Depending on the configuration the system can reliably produce PW or WFI. If a single pass RO is installed, the product will meet all the PW specifications, if a double pass RO configuration is selected, the product will meet all the WFI specifications. A third option is possible, a single pass RO with a downstream UF that will also meet WFI specifications. From experience, even the single pass RO and EDI configuration meet WFI specifications⁽¹⁷⁾ but would not be recognized by the EP as a viable production method as it utilizes only one membrane barrier. The selection of the configuration, RO-EDI-UF or RO-RO-EDI, is to be performed after consideration of the feed water ionic concentrations. A typical system flow diagram for PW production with an ESR-HOD-RO-EDI combination can be seen in Figure 1.
A typical system flow diagram for WFI production with an ESR-HOD-RO-RO-EDI combination can be seen in Figure 2.

The system is completely Stainless Steel (SS) based and is completely hot water sanitizable (HWS). The hot water sanitization cycle is needed for startup or after opening the system for maintenance. Chemicals can be used but the hot water is simple and effective. The water can be heated to over 90°C in the city water tank and due to heat loss will wash through the RO-EDI at below 85°C. In this way all the system will be sanitized, from the city water tank, and up to and including the PW/WFI fill valve on the storage tank. Chemical washes typically need manual intervention for initiation and for post validation that the chemical agent has been rinsed out of the system. The big advantage of the hot water sanitization is that it can be performed automatically during minimum demand periods.
Another advantage of the HWS is that it is much more effective than chemical disinfectants⁽⁸⁾.

Case study:
An ESR-HOD-RO-RO-EDI system has been operating for 2 years producing PW. This PW feed a storage and distribution system that services the production. We will refer to this system as “RO based system”. The PW loop also feeds a Multi Effect (ME) thermal distillation unit for production of WFI. We will refer to this system as “ME based system”. Table 1 summarizes results for the two systems .

As can be seen from table 1, the results for both systems meet the WFI criteria for total micro, endotoxin and nitrates. The Endotoxin was not tested for the RO based system as the system was defined as PW and not WFI. The Endotoxin is not considered to be a problem especially for a double pass RO system⁽¹⁷⁾. The PW system gave higher microbial results as the system is cold as opposed to the WFI that is hot. In addition, more samples were gathered from the PW system and this explains the higher levels of bacteria. Over the full sampling period, both systems constantly met the conductivity (<1.3 µS/cm@25°C) and TOC criteria (both always below 50 ppb) that were measured online.

Production Cost
The cost of production of WFI with thermal process in a ME still varies per site, per cost of energy, per production flow rate and number of distillation columns. The WFI costs can vary from a bare minimum of 12 €/m³⁽¹⁸⁾ to a maximum of 40 €/m³ as calculated by the author. This WFI production price is added on the base price of the RO based system for production of the WFI feed water. The base costs for the RO production of the WFI feed water range from 6 Euro/m³ to 10 Euro/m³. All of these costs are very site specific and change per local fresh water costs, per cost of energy, per the system utilized to generate the WFI and WFI feed water. The typical price for PW production can be estimated at 8 Euro/m³ and the typical price for WFI production can be estimated at 30 Euro/m³.

3. Summary
After many years of deliberation, the EP has changed its standards for WFI production to include membrane based systems.
The main reason for the delay in approving membrane based systems was the difficulty of controlling bacteria in media based systems, softeners and carbon filters. The unchecked growth of the bioburden is liable to lodge in the RO concentrate compartment and to grow through to the permeate. A new system of pretreatment was presented, with ESR and HOD, which its media free and without chemicals. The ESR-HOD combination uniquely controls bacteria by constantly and actively disinfecting the water with chlorine and high doses of UV radiation giving Constant Bacterial Reduction (CBR) without needing down time for sanitization. A case study was presented showing consistent WFI standard results with an ESR-HOD-RO-RO-EDI combination. The WFI ME was fed by this product water and also constantly produced WFI standard water. The additional price of WFI production (installation and unit purchase were not discussed) ranges 12-40 Euro/m³.

4. Discussion
As both the ME based system and the RO based system meet the WFI criteria, we can discuss the advantages and disadvantages of both systems. The main pros and cons are Installation Costs and Operational Costs. The ME based system has to have RO pretreatment as feed water, otherwise the distillation unit will rapidly be blocked by hardness scale. On the other hand this RO pretreatment can be single pass and relatively simple in construction. In our case study this RO pretreatment system also produced PW (to WFI standards as shown by the sampling data in Table 1) and so this RO based system was double pass and was sophisticated. The ME based system was a distillation unit with the PW as feed water.
In our case, the advantages and disadvantages can be summarized in Table 2:

Conclusion
In the case study there was no appreciable difference in performance between the RO based system and the ME based system and EP accepts this system as legitimate for production of WFI.
As there are higher costs for operation of the ME in comparison to the RO based system and the results are the same there is no reason to select RO based systems for generation of cold WFI.”