Document V0e43ZOXoLqoa2ZzMvYmrpQw

T 1 Sterile Filtration Properties of Filter Sheets Made of Cellulose-Asbestos Mixtures and Filter Membranes. Combination of Both Filter Media in Practice. Dipl.-Chem. Dr. Herbert Wilke Process Engineering Laboratory of SEITZ-Werke GmbH, Bad Kreuznach (W. Germany) Abstract Nowadays, both filter media made of celluloseasbestos mixtures (depth filters) as well as filter membranes (sieve filters) are used for sterilizing filtration duties. Both have a number of advantages and disadvantages. Their filtration behaviour is a function of the microstructure which is explained by electron-microscope photographs and model illustrations. Practical examples demonstrate the differences of some technical filtration data such as organism retention, filtering velocities, adsorption behaviour and resistance to clogging. From these data one may chose the optimum filter media. The composition of the liquids, including the nature of impurities and differing viscosities, in many cases necessitate a combination of membrane filters and pad filters to obtain optimum filtration results. I Sterilizing filtration was realized for the first time about 100 years ago through the employment of tight candle type ceramic filters. However, sterilizing filtration of liquids on a commercial scale was not possible before the invention of SEITZ sterilizing sheets made of mixtures of cellulose and asbestos in the years 1913--1914. In the Fifties, this technique was supplemented by finely pored filter mem branes developed around 1910 by Zsigmondy, and first produced in 1929 on an industrial scale by Messrs. Sartorius, Gottingen. Both types of filter media are employed side by side in the constantly widening field of fine filtration, and in particular for sterilizing filtration duties which are the subject of discussion in this paper. The externally very differing structures --robust, thick cardboards on the one hand, and mechanically sensitive and fragile, very thin foils on the other hand possess widely varying internal configuration. These differences in structure are the reason for the differing behaviour during filtration. An explanation of the different filtering properties can be derived from the electron- Fig. 1 SEITZ sterilizing sheet EKS View of the surface layers magnified 1,800 times. microscopic photographs 1 -6 and the model illustration figure 7 elaborated from these. Cellulose-asbestos sheets possess a coarse three dimensional network formed by cellulose fibres with a diameter of approx. 30 urn. It is rendered very close and tight by asbestos fibres, which are fibrillized down to a diameter of 30 nm. At a thick ness of 4--5 mm, the liquid flow has to cover a relatively long distance through the maze of fibres. To truly illustrate the full depth of a SEITZ sheet, photograph 1 would have to be extended to a width of 9 meters. Figure 2 shows the sponge structure of a membrane filter. Microscopically, a filter membrane is by no means a two-dimensional sieve with uniform pore diameters. At the enlargement factor of figure 2, it would possess a penetration depth of approx. 30 cm. Figure 1 already demonstrates that it is not possible to classify SEITZ sheets according to pore dia meters. They are characterized according to filtering effect and filter output. Membranes on the other hand can be characterized on the basis of physically measurable pore sizes. Fig. 2 Surface of a SEITZ Filter membrane with a pore size of 0.45 um -- magnified 2,000 times. 3 The pore sizes stated by the manufacturers, how ever, only represent the mean of a more or less ideal pore size distribution curve. In the retention of small particles, the major dif ference between sheet filters and filter membranes is based on the following criteria: sheet filters trap the particles mechanically in the depth of the filter texture (see figure 4), and ad ditionally by adsorption (see figure 5) on the large specific surface area of the fibres. Filter membranes, on the contrary, have to be selected in such a way that their largest pore diameter is smaller than the particles to be re tained (see figure 6). Since the particles collect on the surface of the filter medium, it is possible to use, with limitations, the term " Sieving effect" . Fig. 3 Typical Pore Distribution Curve of Seitz-Membrane 0.45 .im from a production batch. 4 Fig. 4 Test particles of 1.05 and 1.45 m in the 3-dimensional sieve of a sterilizing sheet magnified 6,200 times. Fig. 6 1.2 urn membrane with test globules of 2.02 um diameter at an enlargement factor of 10,000. Fig. 5 Coli bacterium adsorbed on an asbestos fibril in a SEITZ sterilizing sheet. Magnified E. opt. 28,000 times. The most important structural characteristics of filter sheets and membranes for sterilizing filtration duties are illustrated as simplified geometric models in fig. 7. The relative sizes in the illustration serve as an indication of the different behaviour of these filter media during filtration. 5 Filter Pads Seitz-Entkeimungschicht EKS Seitz sterilising sheet EKS Filter Membranes Seitz-Membranfilter 0.22 pm Seitz membrane filter 0.22 pm Struktur Structure Porendurchmesser Pore-diameter Teilchenabscheidung Retention of particles 3-dimensionales Raumnetz aus Zellstoff-Asbest 3-dimensional network of cellulose and asbestos o (1.2 - 1.5 pm) fiktiv representative sterisch-mechanisch und adsorptiv im Filtergefuge steric mechanical and adsorptive within the pad 'ftfpa''1 2-dimensionales ,,Sieb" mit Schwammstruktur aus Hochpolymeren 2-dimensional "sieve" with sponge structure of plastic r- ---------------------- \ a M 0.22 pm _______________________ Absiebung nach Teilchen gre auf der Oberflche sieve-effect on the surface Poren O' : Porenlnge Pore-diameter : Pore-Iength I1 $1 150 pm 'S 300 pm 1Af A 1 2 pm 0.22 pm Ratio EKS : 0,22 /im Membr. ( ^ _gggg 1 : 700 40 : 1 L _ ------------------------------------------------------------------------------- Porenanteil Void volume 85/ Effektives Hohlraum volumen -- Kapazitt fr Feststoffe Total void volume = turbid capacity Ratio EKS . 0.22 pm Membr. 35 : 1 Innere Oberflche Internal surface area Ratio EKS : 0.22 pm Membr. 80 . 1 cmVcm3 Filtersurface D ~ 100 cmVcm2 Filtersurface Fig. 7 Structural characteristics of steriiizing filter sheets and filter membranes 0.22 (im in simplified geometric models. 6 General Characteristics Specific Terms a Effectiveness of Filter b Specific filtration rate c Throughput, total d Adsorption ability e Purity Reduction of organisms [%] [m l/cm 2 min] or [l/m 2 h] at Ap bar [m l/cm 2] or [l/m 2] Effective Substances Pyrogens, Aflatoxin Loss of soluble components f Structural coheseveness Loss of fibers g Anisotropism h Sterilizability Solvent Resistant effects on general characteristics of point a, b, c Resistance to steam a t ------ C Swelling and Solution Fig. 8 Important properties of filter media for sterilizing filtration. Filter Pads Membranes (100%) x 100% xx + +++ +++ + ++ prewashing may be necessary Final Filtration may be required 0 no mistake possible O slight 0 + skin-effect +++ + Some of the most important filtering properties for sterilizing filtration applications are summarized in table figure 8. In the following, the individual characteristics (a) to (h) are briefly described: a) Filtering effect Filter sheets: In aqueous solutions containing low molecular weight components, bacteria retention is very effective. Retention capacities of 1010organisms/ cm2of filter area are possible. In case of virus contamination it depends on particle size 106 (polio) to 1010 (Newcastle) (see fig. 9). Unlike bacteria, retention of virus depends greatly upon the composition of the solution, and in particular, on the presence of detergents, proteins, and other high molecular weight materials (see fig. 10). In practice, filter sheets offer "self protection" against break-through in that the impurities present cause clogging before the bacteria capacity of the filter is exceeded. Filter membranes: Membranes, with appropriate pore sizes adapted to the micro-organisms permit reliable sterilization without influence of the solution on the sterilizing effect. However, break-through of individual organisms has been observed at high organism counts > 107/ml. (see also in fig. 3 the proportion of pores which exceeds the stated mean pore size). b) Filtration rate The specific filtration rate, i.e. the filtrate volume obtainable per time unit and surface area unit can only be determined with absolutely pure, pre filtered liquids. It is a function of filtration pressure and viscosity (see figure 11). 7 Kein Durchbruch beobachtet No breakthrough observed Durchbruch Breakthrough c = Keimzahl pro ml Testlosung Number of organisms per ml test solution 28 45 200 & in nm Fig. 9 Organism retention capacity of SEITZ filter sheets. According to figure 11, filter membranes exhibit a more than 10-fold higher filtration rate compared with filter sheets under otherwise equal conditions. However, this advantage is confined to largely clean unfiltered solutions. c) Total filter output This figure can only be gained empirically, and in dicates the total filtrate quantity obtainable per area unit. It is also a function of the clogging rate, an important factor in the course of filtration. The total filter output is therefore governed by the impurities in the liquid. Naturally, under otherwise equal conditions, the total throughput of mem branes, with their typical surface filtration is less 8 A 1:1 000 diluted broth without additives B 1:1 000 diluted broth +0.1 % peptone C 1:1 000 diluted broth + 0.2 % peptone D 1:1 000 diluted broth + 0.2 % lactalbumin E 1:1 000 diluted broth + 0.01% agar-agar Fig. 10 The effect of proteins on the retention of Coli-T 3-phages in SEITZ-EKS pads. Flow rate ml/h 100 cm2 Pure prefiltered liquids of 20 C at A p = 1 water 2 dialysis concentrate 19,1% 3 human albumin 5% 4 glucose 40% 5 human albumin 20% 6 sesame oil 7 glycerin 80% Fig. 11 Filtering velocity as a function of viscosity in sterilizing sheets and 0.22 um membranes. 9 than that of filter sheets. This is because the sheets are capable of internally trapping at least 30 times more finely divided turbid matter per area unit, a significant advantage of depth filtration. The filtration curves figures 15 to 17 clearly indicate the large differences in the practical behaviour of both filter media. Therefore filter membranes are usually protected against premature clogging by the incorporation of suitable prefilters. d) Adsorption capacity Losses of solution components through adsorption are hardly observed with filter membranes because their mass represents only 4 % of that of filter sheets (50 g/m 2as opposed to approx. 1,300 g/m 2) and in particular because of the considerably smaller " internal surface area" (see Fig. 7). This factor can be of great advantage in the filtration of solutions with a very low concentration of effective sub stances. The high adsorption power of cellulose-asbestos sheets has in most cases no effect on ions or com pounds with molecular weights < 1,000, especially if the concentration of these substances exceeds the figure 0.01 %. The adsorption equilibrium is then reached after a small liquid volume has passed through. The retention capacity of bacteria toxins and myco toxins is, however, considerable, and is of particular importance in practical pharmaceutical applica tions. The ability to eliminate pyrogenic lipo polysaccharides even from such extreme dilutions as 10 8g/ml makes SEITZ EKS sheets the pref erable filter medium for the preparation of paren teral solutions. Several researchers have deter mined that 1 m2of SEITZ EKS sheets are capable of retaining approx. 500 mg of dissolved, genuine coli-pyrogenes from dilute aqueous solutions. How ever, the pyrogen adsorption depends on the com position of the solutions containing high molecular weight components, e.g. proteins are not completely depyrogenized because an adsorption displace ment takes place. Pyrogenes are not retained by 0.22 |im sterilizing membranes. According to the molecular weight of many lipopolysaccharides of approx. 1,000,000, ultra filters with correspondingly small pores would have to be used. This is out of question for 10 practical applications because of the low filter output. e) Purity Filter membranes, exclusively made of pure chemicals, contain only traces of soluble impurities. Softeners or wetting agents present in quantities of 10--20 mg/100 cm2can be removed by washing with 50 ml/100 cm2of H20. Filter sheets on the other hand are completely com posed of natural materials and may contain a number of alien substances which may have a bearing on the filtrate. In most cases a negative effect on the preparations can be completely eliminated by rinsing according to instructions, discarding of the first filtrate quantity, or returning of same into a relatively large unfiltered preparation volume. It may be necessary to pretreat filter sheets with EDTA or an acid rinse to reduce the presence of foreign ions below detectable limits. Rinsing removes the following quantity of ions per m2filter area: Solvent Ca-Mg-Fe--Cu Ni-- H20 20 C 100 mg/m2 90 mg/m2 5 mg/m2 2 mg/m2 1 mg/m2 Tartaric acid 0,5 % + citric acid 0,5 % 630 mg/m2 450 mg/m2 40 mg/m2 13 mg/m2 4 mg/m2 The above mentioned rinse with 50--150 litres/m2 reduces these figures to non-interfering concen trations. f) Structural strength For the preparation of pure and ultra-pure solutions it is important that the filter medium does not release any insoluble impurities into the filtrate. Membranes consist of homogeneous, coherent materials having only traces of particles of a dusty nature on their surfaces. Provided suitable working techniques, membrane filtrates can be regarded as particle-free even in the sub-microscopic range. For this reason they are preferably used for fibre retention duties down stream of other filter systems. It should be men tioned in this context that 3 .un membranes which Fig. 12 SEITZ hardened surface treatment of sheets, demonstrated by cracked areas when using perforated plate supports a) without fibre protection b) with protective layer. will fairly reliably retain particles > 1 are in most cases regarded as adequate for the prepara tion of fibre-free parenteral solutions. However, practical experience indicates that 10--12 im membranes suffice for injection solutions for the elimination of all interfering dust particles, visible with the naked eye. Cellulose-asbestos sheets without protective fibre impregnation are not suitable for the preparation of pure finished products, because, depending on the support of the sheet in the filter unit, varying quantities of large cellulose fragments and asbestos fibres can become detached during the passage of the liquid through the filter medium. For this reason SEITZ filter sheets are provided with a fibre protective layer on one side. This porous film on the outlet side consist of plastic materials which reduces the detachment of fibres longer than a few im. Figure 12 shows the stabilizing effect of the SEITZ impregnation. Although this impregnation is adequate for most uses, additional fibre catcher devices are necessary to prevent the contamination of injectables. At filtering rates of 500 liters/m2, and without fibre catchers, the following is encountered after pre rinsing: 1-- 250 liters filtrate/m2 250-- 750 liters filtrate/m2 750-- 1,500 liters filtrate/m2 1,500--15,000 liters filtrate/m2 3x10 8 g/ml 8x10'9 g/ml 2x10'9 g/ml 5 x 1010g/ml Of the above, approx. 1/10 to 1/5 may represent asbestos. According to our own measurements, the proportion of asbestos fragments with a size of ^ 100 nm length and 33 nm diameter is < 1010g/ml, downstream of fibre catchers whereby it is immaterial whether 3.0 or 1.2 .urn membranes have been chosen. To our knowledge, only one publication by Nicholson et al. exists about the asbestos content in parenteral solutions. Amongst 16 preparations of the US pharmaceutical industry examined, they found 6 in which 10 6to 10 7g/ml asbestos could be determined. A considerable more sensitive determination method than that used by us and Nicholson is 11 employed by TBA in Rochdale/England. Its detection limit for chrysotil-asbestos is approxi mately 2 x 1012g/ml. The comparitive figures for human plasma drivtes filtered according to different methods are a definite proof for the untenability of such one-sided, propagandist negations we are now seeing published. After EKS sheets 40 x 40 cm After " Lifeguard" pleated membrane After 20 Cox filter sheets, circular After Millipore filter 0.22 [im 6 x 10'12 g/ml 3 x 10'10 g/ml 3 x 10'11 g/ml 4 x 10'11 g/ml 2 x 1011 g/ml 6 x 10'12 g/ml 6 x 10'11 g/ml 6 x 10'11 g/ml 3 x 1011 g/ml 5 x 10'11 g/ml 1 x 10-11 g/ml The samples were taken in the sequence listed during one filtration run. These traces of asbestos conform to the general " pollution level," which was found to be 10'6to 10 7g/l in well water, tap water and beverages (see Cunningham and Pontefract). g) Anisotropy -- Effects Microscopically, filter sheets with a bonded surface on the outlet side exhibit an unmistakeable difference between inlet and filtrate outlet side (see figure 12). The latter is smoother and naturally possesses finer pores. If the filter sheets are inserted incorrectly into the filter unit, they may clog rapidly with outputs being reduced to half. Furthermore, SEITZ sheets possess a density gradient from the surface towards the impregnated outlet side which offers advantages with regard to the capacity of turbid matter. Properly manufactured filter membranes usually exhibit a good isotropy, i.e. pore size and pore distribution are almost identical on both faces of the membrane. However, some grades of filter membranes on the market have shown single-sided skin formations, which should be noted. 12 Fig. 13 5 [.im asbestos fibres on a 3 filter membrane, quantity 0.1 m g/cm2. Magnified 10,000 times. This skin effect not only influences output (block age); it also causes considerable errors in bac teria counts when using the current membrane methods. The circumstances are explained in figure 14. h) Sterilizability and compatibility with solvents While filter sheets, incorporated into filter units, can be sterilized with steam at max. 150 C, or dry at max. 160 C without structural damage, mem branes are temperature sensitive. Nitrocellulose membranes withstand max. temperatures of 125 C for 15 min. Much higher temperatures and steam resistance figures are quoted by the manufacturers for filter membranes made of other materials. The best way to sterilize membranes is with EtO. Also compatibility against alkalis, acids and organic solvents depends on the material. Celluloseasbestos sheets are inert against all solvents, but they are attacked by strong alkalis and acids. Filter membranes have to be selected carefully on the basis of compatibility tables published by the manufacturers. E-Scan-Aufnahme 5000 x einer handelsblichen Membran 0.22 um E-Scan-Micrograph 5000 x of a commercially available membran 0.22 pm S n 9 i Dl0 O n n n 10 D n o '0 rn r-P n |-|Q n l=l D r f ^ 0 DnnO d D DDnD D J[ Modell -- Model D r--.O D D Q n (--I Dn n n u n ml/cm2 Keimzahl -- Number of colonies Fig. 14 Influence of "skin effect" on finely pored membranes. 13 Practical examples of output comparisons -- Combination of membranes with filter sheets. It was pointed out in section " c" that the holding capacity of the filter medium for turbid matter is the decisive factor in optimizing filtration systems. Blockage of the pores of any filter medium leads to a reduction of the filter output. Filter membranes retaining the solids on the surface are more prone to blockage than the loosely textured depth filters. In practice, the advantages of relatively high initial flow rates of filter membranes are usually offset rapidly unless suitable measures to prevent clogging are taken. Under favourable conditions, the maximum outputs shown in diagram figure 11 can only be maintained for 1--3 ml filtrate per cm2 with customary solutions and single membrane filtration. On the other hand the behaviour of filter sheets is considerably more favourable unless very coarse, slimy, amorphous turbid matter is en countered which will smear and plug the pores on the filter surface. The following diagrams illustrate the importance of prefiltration when membranes are used for final sterilizing filtration in pharmaceutical applications. The curves 15 a to 17 a show the filtrate volumes as a function of filtering time, related to a filter area of 100 cm2and a constant pressure of 1 bar. The designation " 0.22 Am" implies that an unprotected SEITZ membrane with a pore diameter of 0.22 aiti has been tested, the designation " EKS" stands for the filtration performance of SEITZ EKS sheets. The designation " 0.22 am + GI.F'' stands for glass fibre/ 0.22 [Am membrane combination. The " 0.22 + SSK" stands for the combination SEITZ SSK pads with 0.22 |im membranes. In these tests the pads and glass filters were the same size as the membranes and were placed on top of them. In addition, the filtration pattern of an 0.22 |u,m filter membrane is reproduced in each example. Here, larger quantities of liquid were separately subjected to prior purifica tion through SEITZ EKS or a clarifying sheet. This is marked: " pre-filtered with . . . " . 14 ab Zeit -- Time [min] Vollentsalztes Wasser Deionized water Fig. 15 Effect of the Pre-filter on the Membranes Throughput Zeit -- Time [min] Elektrolyt-Konzentrat fr Haemodialyse Haemodialysis electrolytic concentrate Even the most simple filtration task, namely the preparation of sterile, fully demineralized water, free from suspended solids, proves the absolute necessity to protect filter membranes with the aid of coarse-pored pre-filters. At equal filtration con ditions, diagram figure 15 a shows the results downstream of an ion exchange column operated at 30 l/h per liter of resin. Scour and abrasion particles leached from the resin, plug the unpro tected membrane after a throughput of 4 1/100 cm2. Up to the termination of the test run (not shown in the diagram) in which 100 1/100 cm2were filtered, these impurities caused no output reduction of the SEITZ EKS sheets. Glass fibre sheets, generally employed as pre-filters in combination with filter membranes, afford con siderable yield increases, on average by the factor 8. Direct combination of membranes with pronounced loosely textured filters of the type SSK resulted in a linear relationship between filter output and filtration time whereby clogging could not be ob served up to a total run of 100 1/100 cm2. However, the maximum continuous output of the filter membrane can only be achieved if the pre-filter is adapted to the turbid matter present, not only from the structure, but also from a filter area point of view. Filter areas of any desirable size can be accommodated in separate filter units to be incor porated upstream of the membrane filter holder. An equipment arrangement for a double filtration of this kind is shown in figure 18. In the example quoted, the performance of the combined glass fibre mat/membrane filter falls 15 M 0.22 pm pre-filtered with EKS 3 M 0.22 pm pre-filtered with EKS b Zeit -- Time [min] 20 Human-Albumin Fig. 16 Effect of the Pre-filter on the Membranes Throughput Zeit -- Time [min] 5 "A Human-Albumin below that of the straight sheet filtration, at the latest after a throughput of 30 1/100 cm2. The maxi mum efficiency of the membrane can, however, be utilized if an EKS pre-filter with a 5-fold filtering area is incorporated. This factor can be calculated for every preparation from the inclination ratio of the curves " EKS" to " 0.22 |im, EKS pre-filtered" (= 10 I/60 min. : 55 I/60 min.). Diagram 15 b shows even more extreme con ditions. The example is based on a concentrated electrolyte/glucose solution as composed from commercial DAB-7-preparations for haemo dialysis. The example also demonstrates that optimum outputs can be achieved in direct filter sheet/ membrane filter combinations by careful selection of the pre-filter. 16 Since a pyrogen-free filtrate is necessary in this example, pre-filtration with EKS sheets would be carried out under all circumstances. The ratio " EKS" to " 0.22 urn, EKS pre-filtered" indicates that the pre-filter section must be 17 times the size of the membrane filter section. High molecular solutions in larger concentrations can under no circumstances be handled by a single membrane filtration. The example human albumen, diagrams 16 a and b demonstrates this. Due to the higher viscosity, a 20 % solution can only be filtered through EKS at V3of the output of a 5 % solution. Solutions pre-filtered with EKS can, however, be passed through sterilizing membranes with excel lent flow rates. I a Zeit -- Time [min] 0.2 */ Albumin mit 5.2 x 10* Organismen/ml 0.2 Albumin with 5.2 x 10* Organisms/ml Fig. 17 Filtration Rate as a Function of Organism Loading Keimmenge -- number of organisms [10* Keime/cm5] 0.2 Albumin mit 5.2 x 10* Organismen/ml 0.2 "/ Albumin with 5.2 x 10* Organisms/ml The fact that even micro-organisms can cause a complete blockage of filter media is illustrated in diagrams 17 a and b. A sterile pre-filtered 0.2 % albumin solution was stored under unsterile con ditions for 3 days. During this time, 5.2 x 106 un defined organisms had developed per ml. While an 0.22 ^m membrane was already completely blocked after approx. 3 1/100 cm2, and a glass fibre pre-filter doubled that output, EKS sheets produced more than the 5-fold quantity under the same conditions. Diagram 17 b shows the bacteria load per area unit of a sterilizing filter as a function of the filter rate. An 0.22 im membrane retained approx. 1.5 x 108 organisms per cm2, while SEITZ EKS sheets (within the scope of the diagram) trapped 4 times the quantity. However, it has been confirmed re peatedly that the bacteria load can amount to 100-fold at outputs realizable in practice (see also figure 9). These few examples are intended as an indication that conditions in practice are undoubtedly such that satisfactory filtrate quantities can be obtained by placing a carefully chosen pre-filter directly on top of a filter membrane. The optimum procedure for larger outputs on the other hand consists in separate pre-filtration with filter sheets followed by a membrane filter holder designed to produce the desired filtering velocity. 17 Fig. 18 SEITZ Sheet Filter ORION in combination with a SEITZ membrane filter holder EFM 22. 18 Bibliography 1) Altenschmidt, W. und Jaeger, F. "Vergleichen und Beurteilen der Leistung von Sterilfiltern". Pharm. Ind. 33, 72-- 75 (1971) 2) Badenhop, C. T. et al In "Membrane Science and Technology", edited by I. E. Fliun Plenum Press, New York-London 1970, 120-- 138 3) Brock, N. und Geks, Fr. sowie Lorenz, D. "Der Einflu gelster Substanzen auf die Entfernung wasserlslicher Pyrogene durch Filtration (Seitz-Filter)". Arzneim.-Forsch. 5, 475-- 478 (1955) 4) Cunningham, H. M. und Pontefract, R. "Asbestos Fibres in Beverages and Drinking Water". Nature 232, 332-- 333 (1971) 5) Drescher, L. "Filtrationsversuche mit Viren". Dissertation aus Robert Koch Institut, Berlin (1961) 6) Frank, P. "Aflaxtoxine als Gefahrenquelle fr Infusions lsungen". D. Apoth. Zeitung 112, 838-- 841 (1972) 7) Hechmat, A. "Zur Frage der Notversorgung mit Trink wasser mittels Seitz-Filter". Zbl. Bakt. Hyg. I. Abt. Orig. B 155, 535-- 540 (1972) 8) Hechmat, A. "Experimentelle Studie zur Adsorption von Poliomyelitis-Impfvirus Typ 2 bei Filtration durch Millipore-Membranfilter". Bakt. Hyg., I. Abt. Orig. A 219, 420-- 425 (1972) 9) Jordana, R. d. V. "Study on Adsorption of Bacteriophage by Filters". Appl. Microbiol. 7, 239-- 247 (1959) 10) Maier, K. K. und Beutelspacher "Beitrge zur Kenntnis der Hohlraumsysteme von Membranfiltern". Kolloid Z. 135, 10-- 28 (1954) 11) Mares, I., Drevo, D. Slonim, D. "A Study on the Preven tion of Polio-Virus Adsorption on Seitz-Filter Pads". Acta Virol. 5, 220-- 227 (1961) 12) Nagy-Takcsi, G. und Renner, E. "Prfung der ehern. Eigenschaften von Filtern". Gygyszereszet 12, 342-- 347 (1968) 13) Nicholson, W. I., Maggiore, C. I., a. Selikoff, I. J. "Asbestos Contamination of Parenteral Drugs". Science 177, 171-- 173 (1972) 14) Skinner, F. S. "Verunreinigungen parenteraler Lsungen durch Fremdteilchen -- ihre Erkennung und Vermeidung". Pharm. Ind. 33, 677-- 684 (1971) 15) Smith, D. K. "Microbiological Test with 0.45 xm Membranes". Ontario Research Foundation June 29, (1970) 16) Sprockhoff, H. v. und Waitz, H. "Untersuchungen ber die Filtration von bakterien- und virushaltigem Wasser mit Seitz-Entkeimungsschichten". Zentral Bl. Vet. Med. B 12, 231-- 240 (1965) 17) Steinke, G. "ber Untersuchungen an Entkeimungs filtern". D. Apoth. Zeitung 108, 1322-- 1326 (1968) 18) Trger, H. "Ultrareinigung und Entkeimung mit Membranfiltern". Verf. Techn. 4, 465-- 470 (1970) 19) Uecker, W. "Membranfilter in der Bakteriophagenfiltration". Zbl. Bakt., I Orig. 168, 194-- 203 (1957) 20) Voggel, K. H. und Perl, H. W. "Zur Sterilfiltration und Sterilittskontrolle pharmazeutischer Produkte mit Membranfiltern". Pharm. Ind. 32, 474-- 478 (1970) 21) Wallhuer, K. H. und Schmidt, H. "Sterilisation-- Desinfektion". Georg Thieme Verl. Stuttgart (1967) 22) Wendland, W. "Filters for Removing Bacteria and Pyrogens from Pharmaceutical Injection Solutions". Filtr. Separ. 446-- 447 (1967) 23) Wilke, H. und Albrecht, I. "Adsorptionsfiltration von Viren mit Seitz-Entkeimungsfiltern". Archiv fr Hygiene und Bakteriologie 145, 2 (1961); 146, 1 (1962) 24) Wilke, H. und Voss, H. E. "Die Entfernung pyrogener Substanzen aus Injektionslsungen durch Filtration". Arzneim.-Forsch. 4, 8-- 14 (1954) 25) Wilke, H. "Herstellung von Konzentraten fr die extracorporale und peritoneale Dialyse". Med. Technik, Heft 1 (1973) SEITZ-Werke GmbH D-6550 Bad Kreuznach Telephone (06 71) 21 01 Telex 4-2851 seitz d _II_I__IN__I_I_I_I_ SEITZ-Filter-Werke Theo & Geo Seitz D-6550 Bad Kreuznach Telephone (06 71) 22 26 Telex 4-2828 sfw bk