FILTERS AND FILTRATION

Published in Encyclopedia of Pharmaceutical Technology

ISBN: 0-8247-2826-2

Maik W. Jornitz 1
1
Sartorius Group
Germany

Encyclopedia Entry | Print Published: 05/22/2002 | Online Published: 01/23/2002

Pages: 1237 - 1249 | PDF File Size: 380 KB
DOI: 10.1081/E-EPT-100001057

Introduction
The separative process of filtration is widely used within the biopharmaceutical
industry to remove contaminants from liquids, air, and gases, such as particulate matter but
especially microorganisms. Microorganism removal is either required to achieve a sterile
filtrate or, if the pharmaceutical product is thermally sterilized, to reduce the bioburden and,
therefore, avoid elevated levels of endotoxins—the debris of gram-negative organisms (1) .

There are many filter configurations within the industry, such as sheet or modular
depth filter types for prefiltration purposes, flat filter membranes mainly for microbial
detection and,specifications, and, most commonly, filter cartridges containing either depth
filter fleeces or membrane filters. Such membrane filters are available in a large variety of
membrane polymers for different applications. These materials are discussed later in the
chapter.

Sterilizing grade membrane filters are defined by the FDA Guideline on Sterile Drug
Products Produced by Aseptic Processing (2) by being able to retain 10 Brevundimonas
7

diminuta (formerly Pseudomonas diminuta) organisms per square centimeter of filtration area
at a differential pressure of 2 bar (3) . Such retention efficiency has to be validated, using the
actual drug product and the process parameters, due to the possibility of an effect to the filters
compatibility and stability and/or the microorganisms size and survival rate (4-5) . Performing
these so-called product bacteria challenge tests became a regulatory demand (6) and,
therefore, belong to a standard filter validation. Before these challenge tests can be performed,
several parameters, for example, bactericidal effects of the product, have to be evaluated. The
recently published PDA Technical Report 26 (7) describes the individual parameters—the
possible effects and mechanisms to be used to perform challenge tests. Additionally, the
report discusses filtration modes, sterilization, and integrity testing.
Filter Types
One can differentiate filters in different distinctive types, commonly in membrane and
depth filters. Depth filters retain contaminants within the depth of the filter matrix.
Contaminants have to move through the tortuous path of the fiber matrix and eventually will
collide with a fiber and separate from the medium. Due to the depth retention, such filters
have a very high dirt-load capacity and are able to separate a high load of contaminants of
different sizes. Depth filters are utilized for coarse particle removal, polishing filtration, and,
especially, to protect final membrane filters reverse osmosis or deionizing units. Depth filters
can greatly enhance the membrane filter's total throughput capability. Therefore, before
utilizing a filtration process in a new application, filterability trials are commonly performed
to evaluate the optimal prefilter–final filter combination to achieve the lowest cost per liter
ratio and highest yield. Initial filterability tests are done with 47-mm composites (flat filter
discs of the filter cartridge device to be used). Having found the optimal combination of
prefilter retention to final filters pore size, pleated small-scale devices are used to achieve
appropriate filter sizing parameters. Filterability trials are performed with automatic test rigs,
which utilize a balance as a load cell to measure the filtrate collected. Commonly, the balance
is connected to a computer system, which uses a specific software showing the flow rate, total
throughput, and differential pressure graphs and offer a report including such.

As depth filters retain contaminants within the fiber matrix, membrane filters are
surface retentive filters and, therefore, have the distinct disadvantage to clog faster. The filter
industry, therefore, pleats such membranes to install a higher effective filtration area into a
filter device. Still, such effort has its limitation, due to a maximum allowable pleat density.
Having reached the limit, the only option is the use of prefilters or membranes of different
pore sizes to gain a fractionate retention and, therefore, a prolonged lifetime of the filter.
Some membrane filter configurations have such membrane or depth filter prefilters built into
the filter cartridge. This is convenient for the filter user in respect to lowered hardware costs.
Additional prefilter housings are not necessary. In comparison to depth filters, membrane
filters have a narrow pore size distribution, which results in a by far sharper retention rate.
Pore size ratings are facilitated to differentiate membrane filters and the performance of such.
Commonly, a sterilizing grade filter is labeled 0.2 m when it retains 107 B.diminuta/cm2.
Another advantage and necessity of membrane filters is the fact that these are integrity
testable. Therefore, flaws or defects can be detected, which is critical, due to the function of
membrane filters, mainly in separating microorganisms from pharmaceutical solutions.

Membrane filters are made in a wide variety of pore sizes (Fig. 1 ). The effective pore
size for membranes vary, and membranes can be used in reverse osmosis (RO), nanofiltration
(NF), ultrafiltration (UF), and microfiltration (MF). RO membranes are widely used in water
treatment to remove ionic contaminations from the water. These membranes have an extreme
small pore size and, therefore, require excellent pretreatment steps to reduce any fouling or
scaling of the membrane, which would reduce the service lifetime. RO membranes are used
by extensive pressures on the upstream side of the filter membrane to force the liquids
through the pores.
 Fig. 1 Typical pore sizes for membranes used in reverse osmosis, nanofiltration, ultrafiltration, and
microfiltration.

The retention ratings of UF filters are also not measured in pore size but rather in
MWCO (molecular weight cut-off), i.e., the molecular weight of the substance to be retained.
UF filter systems are most often used in cross-flow (tangential flow) systems. The feed stream
is directed over the actual membrane to diminish blockage of the membrane. Depending on
the pressure conditions, the fluid (permeate) penetrates through the membrane, whereby the
remaining fluid is recirculated (retentate). UF filter systems find applications in concentration,
diafiltration, and removal steps within pharmaceutical downstream processing. MF can be
used as dead-end filtration (the feed is directed to the membrane, resulting into a filtrate,
separated from the contaminant) or tangential flow mode. The tangential flow characteristic
for MF is commonly used for cell or cell debris removal in downstream processing. MF
membranes typically differ from UF membranes in the morphology of the membrane's cross-
sectional cut. The symmetry of sterilizing-grade microfilters usually ranges from being
uniform to being slightly asymmetric. Ultrafilters, on the other hand, are highly asymmetric,
with the rejecting layer consisting of a tight skin (0.5–10 to m thick) supported by a thick
spongy structure of a much larger pore size (Fig. 2 ).

Fig. 2 Skin layer structure of a UF membrane. (Courtesy of Sartorius AG.)

MF is used in a large variety of filtration applications, from fine cut prefiltration to

sterilizing grade filtration in aseptic processing. Often sterilizing grade filters are the terminal
step before filling or final processing of the drug product. MF is available for air and gas
applications and liquid clarification or sterilization. For the different applications, specific
membrane configurations and materials have been developed.

Filter Materials
There are a variety of different depth filter and membrane filter materials used within
the pharmaceutical processes. Depth filter are fibrous materials: for example, Polypropylene,
Borosilicate, or Glassfibre fleeces (Fig. 3 ). Borosilicate and Glassfibre materials are highly
adsorptive and commonly used to remove colloidal substances, like iron oxide from water or
colloidal haze from sugar solutions.

Fig. 3 SEM of the random fiber matrix of a depth filter. (Courtesy of Sartorius AG.)

Prefilters can also contain membranes instead of fibrous depth filter material. Such
membrane material are commonly mixesters of cellulose or pure cellulose acetate. The
cellulose mixester filter material contains a high degree of cellulose nitrate, which again is
highly adsorptive. Such prefilters have a very sharp retention rating and, therefore, are used in
applications in which the contaminant has a narrow size distribution and/or a sterilizing grade
filter has to be protected, due to the fact that such a filter cannot be changed during the
filtration process. Membrane prefilters, when blocked, can be exchanged during the filtration
process, due to the final filter downstream.

Most commercial UF and NF membranes and many MF membranes are made by the
phase-inversion process, where a polymer is dissolved in an appropriate solvent along with
appropriate pore-forming chemical agents. The polymer solution is cast into a film, either on a
backing or freestanding, and then the film is immersed in a nonsolvent solution that causes
precipitation of the polymer. Such membranes are Polyamides, such as Nylon,
Polyethersulfon (PESU), or Polyvinyldienefluoride (PVDF). Cellulosic membranes, such as
cellulose nitrate, acetate, or regenerated cellulose, are casted as a cellulose–water–solvent
mixture onto a belt and transported through heated tunnels. The resulting evaporation process
produces the porous structure of the membrane seen in Fig. 4 .
 Fig. 4 Porous structure of Celluloseacetate. (Courtesy of Sartorius AG.)

Other techniques for membrane formation include stretching the polymeric film,
commonly Polytetrafluoroethylene (PTFE), while it is still in a flexible state and then
annealing the membrane to “lock in” and strengthen the pores in the stretched membrane. The
stretching process results into a distinctive membrane structure of PTFE nodes, which are
interconnected by fibrils, (Fig. 5 ).

Fig. 5 PTFE membrane structure. (Courtesy of Sartorius AG.)

 PTFE membranes are highly hydrophobic and, therefore, are used as air filters. Air
filters have to be highly hydrophobic to avoid water blockage due to moisture or condensate,
especially after steam sterilization of these filters. Water blockage could be detrimental, if the
filter is, for example, used in a tank venting application to overcome condensation vacuum of
a nonvacuum resistant tank. If the filter would not allow a free flow of air into the tank, it may
implode. Therefore, vent filters for this application have to be chosen and sized with care.
PTFE membranes are also highly mechanical and thermal resistant, which is required,
because such filters are used over several months, withstanding multiple steam-sterilization
cycles. Especially in large-scale fermentation, these filters are used over several months,
avoiding unwanted infections of the fermenter's or bioreactor's cell line.

Finally, track-etched MF membranes are made from polymers, such as polycarbonate

and polyester, wherein electrons are bombarded onto the polymeric surface. This
bombardment results in “sensitized tracks,” where chemical bonds in the polymeric backbone
are broken. Subsequently, the irradiated film is placed in an etching bath (such as a basic
solution), in which the damaged polymer in the tracks is preferentially etched from the film,
thereby forming cylindrical pores. The residence time in the irradiator determines pore
density, and residence time in the etching bath determines pore size. Membranes made by this
process generally have cylindrical pores with very narrow pore-size distribution, albeit with
low overall porosity. Furthermore, there always is the risk of a double hit, i.e., the etched pore
becomes wider and could result in particulate penetration. Such filter membranes are often
used in the electronic industry to filter high-purity water.

Table 1 lists the different membrane polymers available and the advantages and
disadvantages, which depend on the properties of the polymer. The table shows that there is
no such thing as a membrane polymer for every application. Therefore, filter membranes and
the filter performance have to be tested before choosing the appropriate filter element.

Table 1 Advantages and disadvantages of various membrane polymers

Membrane
Advantages Disadvantages
material
Very low nonspecific adsorption
(nonfouling) Limited pH compatibility
Cellulose acetate
High flow rates and total Not dry autoclavable
throughputs
Cellulose nitrate Good flow rate and total High nonspecific adsorption
(nitrocellulose) throughputs Limited pH compatibilityNot dry autoclavable
Very low nonspecific adsorption
Regenerated (nonfouling) Limited pH compatibility
cellulose Very high flow rates and total Not dry autoclavable
throughputs
Very low nonspecific adsorption
(nonfouling)
Modified
Moderate flow rates and total
regenerated Ultrafilters not dry autoclavable
throughputs, especially with
cellulose
difficult to filter solutions
Broad pH compatibility
Good solvent compatibility
High nonspecific protein adsorption Low hot-
Good mechanical strength
Nylon 66 water resistance
Broad pH compatibility
Moderate flow rate and total throughput
Dry autoclavable
Moderate flow rates
Polycarbonate Good chemical compatibility Low total throughputs
Difficult to produce
Polyethersulfone High flow rates and total Moderate-to-low nonspecific adsorption,
 Membrane
Advantages Disadvantages
material
throughputs depending on surface modifications.
Broad pH compatibility Limited solvent compatibility
(Continued )
Hydrophobic material
Excellent chemical resistance
Polypropylene High nonspecific adsorption due to
High mechanical resistance
hydrophobic interactions
High flow rates and total
Moderate-to-high nonspecific adsorption
Polysulfone throughputs
Limited solvent compatibility
Broad pH compatibility
Hydrophobic material
Polytetrafluoro- Excellent chemical resistance High nonspecific adsorption due to
ethylene High mechanical resistance hydrophobic interactions
High-cost filter material
Moderate flow rate and total throughput
Hydrophobic base, made hydrophilic by
Low nonspecific adsorption
Polyvinylidene- chemical surfacetreatment; may lose
Dry autoclavable
difluoride hydrophilic modification due to chemical
Good solvent compatibility
attack
High-cost filter material

Table 1 (Continued)
Excellent chemical Hydrophobic material
resistance High nonspecific adsorption due to hydrophobic
Polytetrafluoro-
High mechanical interactions
ethylene
resistance High-cost filter material
Low nonspecific Moderate flow rate and total throughput
adsorption Hydrophobic base, made hydrophilic by chemical surface
Polyvinylidene-
Dry autoclavable treatment; may lose hydrophilic modification due to
difluoride
Good solvent chemical attack
compatibility High-cost filter material

Filter Construction
Filters are available in several constructions, effective filtration areas, and
configurations. Depending on the individual process, the filter construction and setup will be
chosen to fit its purpose best. Most commonly used for RO filters are tubular devices, so-
called spiral wound modules due to the spiral configuration of the membrane within the
support construction of such device. UF systems can be found as a spiral wound module, a
hollow fiber, or a cassette device. The choice of the individual construction depends on the
requirements and purposes towards the UF device. Similar to the different membrane
materials, UF device construction has to be evaluated in the specific applications to reach an
optimal functioning of the unit. Microfilters and depth filters can be lenticular modules or
sheets but are mainly cylindrical filter elements of various sizes and filtration areas, from very
small scale of 300 cm2 to large scale devices of 36 m2. A 10-inch high cylindrical filter
element can be seen in Fig. 6 .
 Fig. 6 10-inch standard filter element with pleated membrane and protection fleeces. (Courtesy of
Sartorius AG.)

These filter elements are installed into stainless steel filter housings by pushing the
double O-ring cartridge adapter into the housing base plate recess. The filter housing is then
assembled and connected, and the filter is flushed with water and steam-sterilized, either by
in-line steaming or autoclaved. If filter housings are not available or not preferred, disposable
filters can be used. The filter element is welded into a plastic housing, usually Polypropylene,
and after every filtration process, discarded. The advantage of such a disposable filter device
is the reduced cleaning validation effort, and the user does not come in contact with the
filtered product. Such disposable filters can be autoclaved, but not in-line steam-sterilized,
due to the pressure–temperature ratio of the housing polymer. Most often, such disposable
filters are used for scale-up filtration tests, due to the ease of use and the availability of a band
of effective filtration areas.

Filter Validation
Pharmaceutical processes are validated processes to assure a reproducible product
within set specifications. Equally important is the validation of the filters used within the
process, especially the sterilizing grade filters, which, often enough, are used before filling or
the final processing of the drug product. In its Guideline on General Principles of Process
Validation, 1985 (8) , and Guideline on Sterile Drug Products Produced by Aseptic
Processing, 1987 (2) , the FDA makes plain that the validation of sterile processes is required
by the manufacturers of sterile products.

Sterilizing grade filters are determined by the bacteria challenge test. This test is
performed under strict parameters and a defined solution (ASTM F 838-83) (3) . In any case,
the FDA nowadays also requires evidence that the sterilizing grade filter will create a sterile
filtration, no matter the process, fluid or bioburden, found (6) , (9) . This means that bacteria
challenge tests have to be performed with the actual drug product, bioburden, if different or
known to be smaller than B. diminuta and the process parameters. The reason for the
requirement of a product bacteria challenge test is threefold. First, the influence of the product
and process parameters to the microorganism has to be tested. There may be cases of either
shrinkage of organisms due to a higher osmolarity of the product or prolonged processing
times. Second, the filter's compatibility with the product and the parameters has to be tested.
The filter should not show any sign of degradation due to the product filtered. Additionally,
rest assurance is required that the filter used will withstand the process parameters; e.g.,
pressure pulses, if happening, should not influence the filter's performance. Third, there are
two separation mechanisms involved in liquid filtration: sieve retention and retention by
adsorptive sequestration (1) , (8) , (10-12) . In sieve retention, the smallest particle or
organism size is retained by the biggest pore within the membrane structure. The contaminant
will be retained, no matter the process parameters. This is the ideal. Retention by adsorptive
sequestration depends on the filtration conditions. Contaminants smaller than the actual pore
size penetrate such and may be captured by adsorptive attachment to the pore wall. This effect
is enhanced using highly adsorptive filter materials, for example, Glassfibre as a prefilter or
Polyamide as a membrane. Nevertheless, certain liquid properties can minimize the adsorptive
effect, which could mean penetration of organisms. Whether the fluid has such properties and
will lower the effect of adsorptive sequestration and may eventually cause penetration has to
be evaluated in specific product bacteria challenge tests. Table 2 shows the advantages and
disadvantage of both separation mechanisms.

Table 2 Advantanges and Disadvantages of Separation Mechanisms

Retention
Advantages Disadvantages
mechanism
Reliable at worst case product properties
Reliable separation even at high flows and
pressure conditions
Blockage, i.e., exhaustion, can be Retentive only at the specific pore
Sieve retention
determined size rating
No unspecific adsorption, minimal loss of
desired product, and little adsorptive
fouling
Highly influenced by product specific
properties
Separated particles can be shed by
It is possible to retain particles smaller than
varying process conditions
the filter's indicated pore size
Adsorptive Saturation of the active sites cannot
Separation of colloidal substances is
sequestration be determined, no warning
possible
Unspecific adsorption will result in
In some case, pyrogens can be removed
product losses and fouling
Lower reliability in terms of absolute
separation

Before performing a product bacteria challenge test, it has to be assured that the liquid
product does not have any detrimental, bactericidal or bacteriostatic, effects on the challenge
organisms. This is done utilizing viability tests. The organism is inoculated into the product to
be filtered at a certain bioburden level. At specified times, the log value of this bioburden is
tested. If the bioburden is reduced due to the fluid properties, a different bacteria challenge
test mode becomes applicable (7) . If the reduction is a slow process, the challenge test will be
performed with a higher bioburden, bearing in mind that the challenge level has to reach 107
per square centimeter at the end of the processing time. If the mortality rate is too high, the
toxic substance is either removed or product properties are changed. This challenge fluid is
called a placebo. Another methodology would circulate the fluid product through the filter at
the specific process parameters as long as the actual processing time would be. Afterwards,
the filter is flushed extensively with water and the challenge test, as described in ASTM
F838-38, performed. Nevertheless, such a challenge test procedure would be more or less a
filter compatibility test.

Besides the product bacteria challenge test, tests of extractable substances or

particulate releases have to be performed (7-8) , (13) . Extractable measurements and the
resulting data are available from filter manufacturers for the individual filters. Nevertheless,
depending on the process conditions and the solvents used, explicit extractable tests have to
be performed. These tests are commonly done only with the solvent used with the drug
product but not with the drug ingredients themselves, because the drug product usually covers
any extractables during measurement. Such tests are conducted by the validation services of
the filter manufacturers using sophisticated separation and detection methodologies, as GC-
MS, FTIR, and RP-HPLC. These methodologies are required, due to the fact that the
individual components possibly released from the filter have to be identified and quantified.
Elaborate studies, performed by filter manufacturers, showed that there is neither a release of
high quantities of extractables (the range is ppb to max ppm per 10-inch element) nor have
toxic substances been found (13) .

Particulates are critical in sterile filtration, specifically of injectables. The USP 24

(United States Pharmacopoeia) and BP (British Pharmacopoeia) quote specific limits of
particulate level contaminations for defined particle sizes. These limits have to be kept and,
therefore, the particulate release of sterilizing grade filters has to meet these requirements.
Filters are routinely tested by evaluating the filtrate with laser particle counters. Such tests are
also performed with the actual product under process conditions to proove that the product,
but especially process conditions, do not result in an increased level of particulates within the
filtrate.

Additionally, with certain products, loss of yield or product ingredients due to

adsorption shall be determined (14-15) . For example, preservatives, like
benzalkoniumchloride or chlorhexadine, can be adsorbed by specific filter membranes. Such
membranes need to be saturated by the preservative to avoid preservative loss within the
actual product. This preservative loss, e.g., in contact lens solutions, can be detrimental, due
to long-term use of such solutions. Similarly, problematic would be the adsorption of required
proteins within a biological solution. To optimize the yield of such proteins within an
application, adsorption trials have to be performed to find the optimal membrane material and
filter construction.

Cases that use the actual product as a wetting agent to perform integrity tests require
the evaluation of product integrity test limits (7) , (17) . The product can have an influence on
the measured integrity test values due to surface tension, or solubility. A lower surface
tension, for example, would shift the bubble point value to a lower pressure and could result
in a false negative test. The solubility of gas into the product could be reduced, which could
result in false positive diffusive flow tests. Therefore, a correlation of the product as a wetting
agent to the, water wet values has to be done, according to standards set by the manufacturer
of the filter. This correlation is carried out by using a minimum of three filters of three filter
lots. Depending on the product and its variability, one or three product lots are used to
perform the correlation. The accuracy of such a correlation is enhanced by automatic integrity
test machines. These test machines measure with highest accuracy and sensitivity and do not
rely on human judgement, as with a manual test (7) . Multipoint diffusion testing offers the
ability to test the filter's performance and, especially, to plot the entire diffusive flow graph
through the bubble point. The individual graphs for a water-wet integrity test can now be
compared to the product wet test and a possible shift evaluated. Furthermore, the multipoint
diffusion test enables the establishment of an improved statistical base to determine the
product wet versus water-wet limits (16-17) .

Filter Integrity Testing

Sterilizing grade filters require testing to assure the filters are integral and fulfill their
purpose. Such filter tests are called integrity tests and are performed before and after the
filtration process. Sterilizing grade filtration would not be admitted to a process if the filter
would not be integrity tested in the course of the process. This fact is also established in
several guidelines, recommending the use of integrity testing, pre- and post-filtration. This is
not only valid for liquid but also for air filters.

Examples of such guidelines are:

1.
FDA Guideline on Sterile Drug Products Produced by Aseptic Processing (1987): Normally,
integrity testing of the filter is performed after the filter unit is assembled and prior to use. More
importantly however, such testing should be conducted after the filter is used in order to detect
any filter leaks or perforations that may have occurred during filtration.
2. The Guide to Inspections of High Purity Water Systems, Guide to Inspections of Lyophilization of
Parenterals, and also the CGMP document 212.721 Filters state the following:

a
The integrity of all air filters shall be verified upon installation and maintained throughout use.
A written testing program adequate to monitor integrity of filters shall be established and
followed. Results shall be recorded and maintained as specified in 212.83.
b
Solution filters shall be sterilized and installed aseptically. The integrity of solution filters shall
be verified by an appropriate test, both prior to any large-volume parenteral solution filtering
operation and at the conclusion of such operation before the filters are discarded. If the filter
assembly fails the test at the conclusion of the filtering operation, all materials filtered through
it during that filtering operation should be rejected. Rejected materials may be refiltered using
filters whose integrity has been verified provided that the additional time required for
refiltration does not result in a total process time that exceeds the limitations specified in
212.111. Results of each test shall be recorded and maintained as required in 212.188(a)
3. ISO 13408-1 First Edition, 1998-08-1, Aseptic Processing of Health Care Products, Part 1: General
requirements: Section 17.11.1 Investigation, m. pre- and post-filter integrity test data, and/or
filter housing assembly:

 20.3.1. A validated physical integrity test of a process filter shall be conducted after use
without disturbing the filter housing assembly. Filter manufacturer's testing instructions
or recommendations may be used as a basis for a validated method. Physical integrity
testing of a process filter should be conducted before use where process conditions
permit. “Diffusive Flow,” “Pressure Hold,” and “Bubble Point” are acceptable physical
integrity tests.

20.3.2. The ability of the filter or housing to maintain integrity in response to sterilization
and gas or liquid flow (including pressure surges and flow variations) shall be determined.
4. USP 23, 1995, P. 1979. Guide to Good Pharmaceutical Manufacturing Practice (Orange Guide,
U.K., 1983):
  PDA (Parenteral Drug Association), Technical Report No. 26, Sterilizing Filtration of Liquids
(March 1998):

Integrity tests, such as the diffusive flow, pressure hold, bubble point, or water
intrusion tests, are nondestructive tests, which are correlated to the destructive bacteria
challenge test with 107/cm2 B. diminuta (1) , (8) . Derived from these challenge tests, specific
integrity test limits are established, which are described and documented within the filter
manufacturers' literature. The limits are water-based; i.e., the integrity test correlations are
performed using water as a wetting medium. If a different wetting fluid, such as a filter or
membrane configuration, is used, the integrity test limits may vary. Integrity test
measurements depend on the surface area of the filter, the polymer of the membrane, the
wetting fluid, the pore size of the membrane, and the gas used to perform the test. Wetting
fluids may have different surface tensions, which can depress or elevate the bubble point
pressure. The use of different test gases may elevate the diffusive gas flow. Therefore,
appropriate filter validation has to be established to determine the appropriate integrity test
limits for the individual process.

Bubble Point Test

Microporous membranes will fill their pores with wetting fluids by imbibing that fluid
in accordance with the laws of capillary rise. The retained fluid can be forced from the filter
pores by air pressure applied from the upstream side. The pressure is increased gradually in
increments. At a certain pressure level, liquid will be forced first from the set of largest pores,
in keeping with the inverse relationship of the applied air pressure P and the diameter of the
pore, d, described in the bubble point equation:

where is the surface tension of the fluid, is the wetting angle, P is the upstream pressure at
which the largest pore will be freed of liquid, and d is the diameter of the largest pore.

When the wetting fluid is expelled from the largest pore, a bulk gas flow will be
detected on the downstream side of the filter system (Fig. 7 ). The bubble point measurement
determines the pore size of the filter membrane, i.e., the larger the pore the lower the bubble
point pressure. Therefore, filter manufacturers specify the bubble point limits as the minimum
allowable bubble point. During an integrity test, the bubble point test has to exceed the set
minimum bubble point.
 Fig. 7 Manual bubble point test set-up. (Reprinted from Technical Report No. 26, Sterilizing Filtration of
Liquids © 1998 by PDA.)

Diffusion Test
A completely wetted filter membrane provides a liquid layer across which, when a
differential pressure is applied, the diffusive airflow occurs in accordance with Fick's law of
diffusion (Fig. 8 ). This pressure is called test pressure and commonly specified at 80% of the
bubble point pressure. In an experimental elucidation of the factors involved in the process,
Reti (18) simplified the integrated form of Fick's law to read as follows:

where N is the permeation rate (moles of gas per unit time), D is the diffusivity of the gas in
the liquid, H is the solubility coefficient of the gas, L is the thickness of liquid in the
membrane (equal to the membrane thickness if the membrane pores are completely filled with
liquid), P (p1-p2) is the differential pressure, and is the void volume of the membrane, its
membrane porosity, commonly around 80%.
 Fig. 8 Manual diffusive flow test set-up. (Reprinted from Technical Report No. 26, Sterilizing Filtration of
Liquids © 1998 by PDA.)

The size of pores only enter indirectly into the equation; in their combination, they
comprise L, the thickness of the liquid layer, the membrane being some 80% porous. The
critical measurement of a flaw is the thickness of the liquid layer. Therefore, a flaw or an
oversized pore would be measured by the thinning of the liquid layer due to the elevated test
pressure on the upstream side. The pore or defect may not be large enough that the bubble
point comes into effect, but the test pressure thins the liquid layer enough to result into an
elevated gas flow. Therefore, filter manufacturers specify the diffusive flow integrity test
limits as maximum allowable diffusion value. The larger the flaw or a combination of flaw,
the higher the diffusive flow.

Pressure Hold Test

The pressure hold test is a variant of the diffusive airflow test (19) . The test set-up is
arranged as in the diffusion test except that when the stipulated applied pressure is reached,
the pressure source is valved off (Fig. 9 ). The decay of pressure within the holder is then
observed as a function of time, using a precision pressure gauge or pressure transducer.
Fig. 9 Manual pressure-hold test set-up. (Reprinted from Technical Report No. 26, Sterilizing Filtration of
Liquids © 1998 by PDA.)

The decrease in pressure can come from two sources: 1)the diffusive loss across the
wetted filter. Because the upstream side pressure in the holder is constant, it decreases
progressively as all the while diffusion takes place through the wetted membrane; and 2) the
source of pressure decay could be a leak of the filter system set-up.

An important influence on the measurement of the pressure hold test is the upstream
air volume within the filter system. This volume has to be determined first to specify the
maximum allowable pressure drop value. The larger the upstream volume, the lower will the
pressure drop be. The smaller the upstream volume, the larger the pressure drop. This also
means an increase in the sensitivity of the test, and also an increase of temperature influences,
if changes occur. Filter manufacturers specify maximum allowable pressure drop values.

Water Intrusion Test

The water intrusion test is used for hydrophobic vent and air membrane filters only
(20-23) . The upstream side of the hydrophobic filter cartridge housing is flooded with water.
The water will not flow through the hydrophobic membrane. Air or nitrogen gas pressure is
then applied to the upstream side of the filter housing above the water level to a defined test
pressure. This is done by way of an automatic integrity tester. A period of pressure
stabilization takes place over time frame, by the filter manufacturer's recommendation, during
which the cartridgepleats adjust their positions under imposed pressures. After the pressure
drop thus occasioned stabilizes, the test time starts, and any further pressure drop in the
upstream pressurized gas volume, as measured by the automatic tester, signifies a beginning
of water intrusion into the largest (hydrophobic) pores, water being incompressible. The
automated integrity tester is sensitive enough to detect the pressure drop. This measured
pressure drop is converted into a measured intrusion value, which is compared to a set
intrusion limit, which has been correlated to the bacteria challenge test. As with the diffusive
flow test, filter manufacturers specify a maximum allowable water intrusion value. Above this
value, a hydrophobic membrane filter is classified as nonintegral.
1
Filtration in the Biopharmaceutical Industry, Meltzer T. H., Jornitz M. W. Marcel Dekker, Inc., New
York, 1998.

2
FDA, Center for Drugs and Biologics and Office of Regulatory Affairs, Guideline on Sterile Drug
Products Produced by Aseptic Processing, 1987.

3
Standard Test Method for Determining Bacterial Retention of Membrane Filters Utilized for Liquid
Filtration, Standard F838-83, American Society for Testing and Materials (ASTM), 1983. Revised
1988.

4
Meltzer T. H., Jornitz M. W., Mittelman M. W., Surrogate Solution Attributes and Use Conditions:
Effects on Bacterial Cell Size and Surface Charges Relevant to Filter Validation Studies, PDA J. Sci.
Technol., 52 (1) , (Jan/Feb 1998)

5
Levy R. V. The Effect of PH, Viscosity, and Additives on the Bacterial Retention of Membrane
Filters Challenged With Pseudomonas Diminuta, Fluid Filtration: Liquid, American Society for
Testing and Materials (ASTM), Washington, DC, 1987.II

6
Validation of Microbial Retention of Sterilizing Filters, PDA Special Scientific Forum, Bethesda, ,
July 1995, pp. 12–13.

7
, Technical Report No. 26, Sterilizing Filtration of Liquids, PDA J. Pharm. Sci. Technol., 52 (S1) ,
(1998)

8
FDA, Center for Drugs and Biologicals and Office of Regulatory Affairs, Guideline on General
Principles of Process Validation, Washington, , September 1985.

9
Jornitz M. W., Meltzer T. H. Sterile Filtration—A Practical Approach, Marcel Dekker, Inc., New York,
2001.

10
Tanny G. B., Strong D. K., Presswood W. G., Meltzer T. H., Adsorptive Retention of Pseudomonas
Diminuta by Membrane Filters, J. Parent. Drug Assoc., 33 (1979) 40–51.

11
Emory S. F., Koga Y., Azuma N., Matsumoto K., The Effects of Surfactant Types and Latex-Particle
Feed Concentration on Membrane Retention, Ultrapure Water, 10 (2) , (1993) 41–44.

12
Osumi M., Yamada N., Toya M., Bacterial Retention Mechanisms of Membrane Filters, PDA J.
Pharm. Sci. Technol., 50 (1) , (1996) 30–34.

13
Reif O. W., Sölkner P., Rupp J., Analysis and Evaluation of Filter Cartridge Extractables for
Validation in Pharmaceutical Downstream Processing, PDA J. Pharm. Sci. Technol., 50 (1996) 399–
410.

14
Hawker J., Hawker L. M., Protein Losses During Sterilization by Filtration, Lab. Practises, 24 (1975)
805–814.

15
Brose D. J., Henricksen G., A Quantitative Analysis of Preservative Adsorption on Microfiltration
Membranes, Pharm. Tech Europe, (1994) 42–49.

16
Waibel P. J., Jornitz M. W., Meltzer T. H., Diffusive Airflow Integrity Testing, PDA J. Pharm. Sci.
Tech., 50 (5) , (1996) 311–316.

17
Jornitz M. W., Brose D. J., Meltzer T. H., Experimental Evaluations of Diffusive Airflow Integrity
Testing, PDA J. Parenter. Sci. Technol., (1998)

18
Reti A. R., An Assessment of Test Criteria in Evaluating the Performance and Integrity of
Sterilizing Filters Bull, J. Parenter. Drug Assoc., 31 (4) , (1977) 187–194.

19
Madsen R. E., Meltzer T. H., An Interpretation of the Pharmaceutical Industry Survey of Current
Sterile Filtration Practices, PDA J. Pharm. Sci. Technol., 52 (6) , (1998) 337–339.
20
Jornitz M. W., Waibel P. J., Meltzer T. H., The Filter Integrity Correlations, Ultrapure Water, (1994
Oct.) 59–63.

21
Meltzer T. H., Jornitz M. W., Waibel P. J., The Hydrophobic Air Filter and the Water Intrusion Test,
Pharm. Tech., 18 (9) , (1994) 76–87.

22
Tarry S. W., Henricksen G., Prashad M., Troeger H., Integrity Testing of EPTFE Membrane Filter
Vents, Ultrapure Water, 10 (8) , (1993) 23–30.

23
Tingley S., Emory S., Walker S., Yamada S., Water-Flow Integrity Testing: A Viable and
Validatable Alternative to Alcohol Testing, Pharm. Tech., 19 (10) , (1995) 138–146.