Andy Feigenwinter, Christian Stark, Christ Water Technology Group
Introduction
Up until a few years ago, sterile and apyrogenic water could only be produced by distillation. Due to rapid progress in membrane technology, and especially in the field of ultrafiltration (UF), an alternative is now available that is as reliable as distillation but offers the additional advantage of being much more economical.
Generalities
Depending on the application, various membrane filtration processes are available. Looking at the various membrane processes, it is obvious that reverse osmosis (RO) shows the lowest cut-off levels (see Table 1).
 |
In general, RO membranes achieve an excellent separation of microorganisms and pyrogens but the module structure shows certain mechanical weaknesses. Such shortcomings are not present in the ultrafiltration modules, which are therefore predestined for the production of high-quality water, such as water for injection (WFI) or highly purified water (HPW).
With their asymmetrical membrane structure, ultrafiltration membranes have no dirt adsorption capacity whatsoever and must therefore be operated in crossflow-there must always be a flow across the membrane surface, which washes out the separated material. An ultrafiltration plant will, for this reason, always produce a concentrate next to the filtrate.
Since fully demineralized water, as it usually circulates in the network of a pharmaceutical facility, often shows a considerable load of germs and organic material, a UF plant must operate with a comparatively high crossflow over the membrane (i.e., the concentrate is recirculated). Furthermore, in most cases back-filtration cycles are scheduled during which the filtrate flows through the membrane in the opposite direction in order to remove deposited material, which is then washed out with the concentrate.
Such back-filtration cycles, which depending on the quality of the demineralized water feed can be necessary in intervals of only a few minutes, use only a fraction of the filtrate produced. Therefore, the filtrate production does not need to be interrupted.
Module structure
Certain requirements are set on the ultrafiltration modules themselves. For example, they must:
- Have a safe cut-off
- Have no short-circuit flow between filtrate and feed water
- Have a high permanent-operation temperature (80°C)
- Be at least hot-water sanitizable (90°C)
- Be preferably steam sterilizable (121°C)
- Have deadspace-free structure (rinsing behavior)
- Allow the possibility for integrity testing
- Allow the possibility for total emptying
Various membrane types are currently available, such as the hollow-fiber module, the spiral-wound module, the plate module, and the tubular module. But even the most perfect membrane becomes useless when the separation between filtrate and feed water is insufficient. This problem is elegantly solved in hollow-fiber modules made of polysulfone (see Fig.1).
 Figure 1. Made of polysulfone, the hollow fiber module has a low cut-off of 6,000 Dalton, far beneath the size of pyrogens of bacterial origin. |
The fiber ends are potted in an epoxy block. Since the epoxy resin reaches up to the permeate nozzle, the requirement for a deadspace-free module structure is fulfilled. O-rings, lip seals, glued and welded joints, which all show a certain risk of leakage, are thus avoided by simple means.
A further advantage of the hollow-fiber modules is based on the fact that the low cut-off of 6,000 Dalton is far beneath the size of pyrogens of bacterial origin. Furthermore, the modules used are structured as double asymmetrical membranes. With two discriminating layers a short circuit is impossible and thus 100 percent safety is ensured.
In order to verify the tightness of the module, an integrity test is installed in ultrafiltration plants, which can be triggered, for example, by a key in the control cabinet. In this verification (pinhole test), the water is emptied from the concentrate side of the modules in order to apply sterile compressed air. If a module is deficient, it is visible by rising air bubbles on the filtrate side. With this simple method, the smallest lesions of the fibers can be discerned by the formation of clearly visible air bubbles.
Since the module housings are transparent, the smallest leakages can thus be detected, the discovery of which would be questionable in the routine endotoxin determination due to the extremely high dilution.
The high separation reliability of such hollow-fiber modules was impressively demonstrated in a challenge test-a test for the retention of bacteria-executed with Hoechst Ltd, recently acquired by sanofi-aventis, in Frankfurt, Germany.
A living culture of more than 107 Pseudomonas diminuta per milliliter was continuously recirculated from a container over a hollow-fiber laboratory module over the course of two weeks. The produced filtrate was fed in its entirety into a sterile filter with a membrane, which was removed once a day and brought to the usual breeding. The system was sterilized only once-before starting the test-with saturated steam at 121°C from the UF module up to and including the sterile membrane. During the test phase the module was loaded with approximately 1013 to 1014 germs. The results showed that not a single germ of the species Pseudomonas diminuta was detected in the filtrate.
Ceramic vs. polysulfone modules
Ceramic modules, which theoretically are an alternative to polysulfone modules, generally show a cut-off of approximately 15,000 Dalton. Ceramic is much more resistant to aggressive chemicals and high temperatures than polysulfone and based on these reasons is mainly applied in wastewater and/or effluent treatment.
The susceptibility to breakage during steam sterilization, the need for seals, and the surface roughness are, however, characteristics that call the use of this module material in the pharmaceutical sphere into question. Furthermore, the rather simple pinhole test cannot be easily executed since air molecules pass the membrane and thus make this verification impossible.
Operation of UF systems
An ultrafiltration system can supply its filtrate in various ways, either into an HPW or WFI storage tank where the pyrogen-free water is then transferred to the consumers by booster pumps. Nowadays, this concept is often used for skid-mounted, compact water systems containing pretreatment, demineralization and ultrafiltration on one single skid.
A second possibility is to take the filtrate directly at the outlet of the UF plant and feed it into the distribution system by means of a pump operated in-line. Both possibilities have advantages and inconveniences. Production into a tank, for example, offers greater flexibility in production and lower investment costs, but has a smaller capacity and requires hot storage (> 65°C) or disinfection of the tank (by ozone, for example) and distribution system. Production into a distribution system does not necessarily require continuous disinfection. The water in circulation is continuously cleaned, and there is higher security against secondary contamination (by tap, for example). However, the plant must be laid out for peak capacity (+20 percent loop return), and a booster pump is required downstream of the UF. Also, there can be a loss of capacity during back-filtration.
 Figure 2. An ultrafiltration plant in hot operation. (Capacity: 50 m3/h) |
A further possibility is the hot operation of a UF plant (see Fig. 2). The temperature has a positive influence on the operating behavior (fouling) of the membranes (see Table 2). Hot operation offers some advantages over cold production: there is constant module capacity, lowering the tendency towards fouling; back-filtration is needed only sporadically; there are longer production cycles between two steam sterilizations; and with hot storage, no further germ-reducing processes (e.g., ozone) are needed.
 |
null
Ultrafiltration vs. reverse osmosis
In principle, the reverse osmosis technique also presents an alternative with its lower nominal cut-off in comparison to ultrafiltration. Certain mechanical shortcomings, such as lip seals and O-rings as separation between raw water and filtrate, require at least the use of two subsequent RO plants.
But since the passage of some germs cannot be reliably excluded even with this system configuration, the subsequent distribution system must in any case (if it is not operated hot) be kept free from germs by the addition of ozone. Reverse osmosis modules are, contrary to UF, not steam sterilizable (though partly hot-water sanitizable) and show a considerably inferior rinsing behavior.
Conclusions
By far the largest quantities of sterile and pyrogen-free WFI-quality water are needed for various rinsing purposes. Depending on the pharmacopeial requirements, WFI-quality water is prescribed, for example, for the final rinsing of vessels for parenterals and also in the final steps of the active substance production for parenteral applications; however, without the production method being stipulated.
The role of ultrafiltration plants (see Fig. 3) for the production of WFI (though allowed by the U.S. and Japanese pharmacopeias) is not predominant because the quantities needed here are usually only a fraction of the water consumption for rinsing purposes. Furthermore, in almost all cases this demand is already satisfied by existing and validated distillation plants.
 Figure 3. An ultrafiltration system with a capacity of 4.5 m3/h. |
Ultrafiltration is outstanding due to its advantages, such as reliability, integrity testing, steamability, and operating cost. From small capacities of 500 l/h and up, ultrafiltration is an extremely economical process and a reliable alternative to classic distillation. For larger capacities, the cost savings multiply due to the low energy cost.
UF operating results show that sterile and pyrogen-free quality is ensured over months and years. Nevertheless, UF plants are sterilized with saturated steam at 121°C at regular intervals. This is done for the purpose of cleaning the membranes only and not to ensure the sterility of the filtrate. Therefore, ultrafiltration is in any case the method of choice for the production of large amounts of WFI-quality water. III
Andy Feigenwinter is sales manager for Christ Ultrapure Water AG, Aesch, Switzerland.
Christian Stark is head of marketing and communications for Christ Water Technology Group, Aesch, Switzerland. He can be reached at [email protected].