Note : Les descriptions sont présentées dans la langue officielle dans laquelle elles ont été soumises.
~6~1~07
SPECI~IC ATION
Microporous membrane sheets are available which have absoiute
particle removal capabilit~T in the range of about 0.1 micron and larger.
These are for the most part made of synthetic resins and cellulose
5 derivatives, and are used as filter media for removing suspended particles
and micro-organisms from fluids.
Such membranes are made using the so-called 1'dry process" by
casting a solvent solution of the resin or cellulose deri~ative on a
temporary support or substrate as a thin film, after which the solvent is
10 removed or exchanged under carefully controlled condItions. Solvent
remo~al and exchange are very slow, and while the process is adaptable
for continuous operation, a very large supporting belt system is required
as the substrate for laydown or casting of the film, and the drying set-up
to carry out removal of the solvent. This Increases plant size and the
15 capital costs in plant construction, and ensures a high cost of manufacture.
Because of the very great length of material (solution or film) which is
in process at any one time, adjustment of processing conditions for close con-
trol of product characteristics is difficult. While the final product is being re-
moved and tested for its characteristics, a very large volume of material is
20 already in process of being formed into a membrane, and past the point where
an adjustment of the process parameters to modify product characteristics,
however prornpt, could affect it. Thus, a considerable amount of out-of-
specification membrane sheet is made before the result of a correction can
be seen at the end of the production line. This results in a large proportion
25 of membrane sheet being out-of-specification, and a wide range of product
.~
07
variation necessarily has to be accepted~ to keep rejec-tions at a minimum.
As a consequence of high production cost and high rejection rate, the price for
such membrane sheet tends to be rather high.
Another process for preparation of membrane sheets also starts
5 from a solution of the resin or cellulose derivative, casting a film of the
solution on a support, and then forming the membrane by precipitation upon
immersion of the film solution Ln a nonso~ent for the resin. This process
results in a skinned membrane, with surface portions having fewer or very
much smaller pores, or even z~ro pores, and an interior portion with larger
10 pores, the outer skinned portions having higher apparent densit~ than the
inter ior por tions .
Sl~inned membranes are nonuniform with respect to particle removal;
for example, the membranes now used for reverse ~smosis are effective in
accomplishing such tasks as g~c or better salt rejection, thus functioning
15 in the 2 to 5 Angstroms (0. 002 to 0. 005 ,uM) range, but are incapable of
providing sterility in the effluent, allowing bacteria in the range of aooo
Angstroms(0. 2 ,uM) to pass. Such membranes are poorly suited when
absol~te removal of particulate material as bacteria is needed.
Thus, for example, Michaels U.S. patentNo. 3,615,024, patented
20 October 26, 19719 describes the formation of anisotropic membranes having
pores of from 1 to 1000 ~lM from a variety of synthetic resins by:
(1) forming a casting dope of a polymer in an o:rganic solvent,
(2) casting a film of said casting dope,
(3) preferentially contacting one side of said film with a diluent
25 characterized by a high degree of miscibility with said organic solvent and a
()7
sufficiently low degree of compatibility with said casting dope to effect rapid
precipitation of said polymer, and
(4) maintaining said diluent in contact with said membrane until
substantially all said solvent has been replaced with said diluent.
The submicroscopically porous a~nisotropic rnembranes consist o~
an integral macroscopically thick film of porous polymer9 usually more
than about 0. 002 and less than about 0. 050 inch in th~ckness~ One
surface of this film is an exceedingly thin but r elatively dense barrier
layer or"skin" of ~om about Q. 1 to 5. 0 microns thickness of microporous
10 polymer in which an aYerage pore diameter is in the millimicron range,
for example from 1. 0 to 1000 millimicrons, i. e., about one-tenth to
one-hundredth the thickness of the skin. The balance of the integral ~ilm
structure is a support layer comprised of a much more coarsely porous
polymer structure through which fluid can pass with little hydraulic
15 resista~ce. By "integral film" is meant continuous, i. e. 9 a continuing
polymer phase. When such a membra~e is employed as a "molecular f-lter"
with the "skin-side" in contact with fluid under pressure, virtually all
resistance to fluid flow through the membrane is encountered in the "skin",
and molecules or particles of dimensions larger thall the pores in the "skin"
20 are selectively retalned. Becallse the skin layer is of such extraordinary
thinness, and because the transition from the skin layer to the macroporous
support structure is so abrupt, normally less than about one-half the thick-
ness of the barrier la~er or less than one micron, the over-all hydraulic
resistance to fluid flow through the membrane i~ very low; that is, the
25 membrane displays surprisingly high permeability to fluids in proportion
to its pore size.
~0'7
Michaels suggests that the formation of these aIlisotropic membranes
appears to be rel~ted to certain diffusional and osrrlotic solvent-exchailge
processes as described hereinbelow:
When a thin layer of polymer solution deposited on a suitable
5 substrate (to assure preferential contact of diluent with orle surface) is
contacted with diluent on one sur-face, diluent and solvent interdiffuse in
the outermost layer almost instantaneously. Thus gelation or precipitation
of the polymer tal~es place almost instantaneously. In view of the rapidity
of this process, the topmost layer of the cast film solidifies as an exceedingly
10 thin membrane skin whose porosity and pore-fineness are governed by the
compatibility criteria developed above. As soon as this membrane skin is
formed, however, the rate of penetration of diluent into the underlying region
of the cast film, and ra~e of extraction of the solvent component, are greatly
retarded. (Itmustnot, however, bestoppedentirely.3 Underthesecircum-
15 stances, subsequent alteration in solution composition within the film occursquite slowly. As a result there i~ opportunity, when a suitable solvent is
present,for slow phase-separation to occur to form a grossly microporous
substructure consisting of large interconnected voids occupiedby solvent/
diluent solution, and an interstitial polymer matrix comprising consolidated,
20 nearly solvent-free polymer. Hence, the formation of a highly permeable,
coarsely microporous substructure is in large part due to proper selection
of a sol~ent system for film-casting dopes arld the selection ~ a proper
diluent for coaction with the solvent system during the precipitation step~
Thus, the Michaels membranes are all skinned, and moreover,
25 while the membranes are water-wettable as long as they are kept wet, once
6~ 7
dried they are all h~drophobic, and difficult to wet with water, except with
the aid of surface-active agents or other wetting aids.
SalemmeU.S. patentNo. 4,032,309, patentedJune28, 1977,
prepares polycarbonate resin membranes described as hydrophobic, evidently
5 of very small pore size, in the ultrafiltration range. Salemme refers to
Michaels U.S. patent No. 3,615,024 and Kimura U.S. patent No. 3,709,7q4,
and st~tes that both Michaels and Kimura utilize the general procedure of
preparing a casting solution of the polymer, casting a film thereof on a smooth
substrate and immersing the substrate and film in an appropriate quenching
10 bath for the development of asymmetric structural cha~acteristics of the
completed film.
These methods differ from each other in the manner in which
some of the process steps are conducted. While the Michaels patent
is particularly directed to the preparation of a membrane having a
15 microporous support layer and an integral microporous skin7 Kimura
is primarily interested in a filrn structure presenting a p~rous region
adjacent a very thin dense nonporous layer. Kimura specifically teaches
the preparation of a casting solution consisting of the polymer and two
mutually miscible solvents in which the polymer is soluble to su~stantially
20 differ2nt degrees. Both the Michaels and Kim-lra methods view the
immersion (or mennbrane-forming) bath as one which Eunctions as a
solvent for the casting solution solvent system, functioning thereby
solely to remove casting solution solvent f~om the film structure.
Contrary to the Kimura process, ~alemme does not employ a
25 three-component (resin, good solvent, poor solvent) casting solution
~ 16~)7
and, in contrast to both Ximura and Michaels, Salemme utilizes an
immersion (quenching) bath to initiate formation of the film that must
provide a function neither disclosed nor contemplated in either Kimura
or ~ichaels; namely, causing swelling of the polycarbonate resin
5 material at the same time as the casting solvent is removed ~om the
film thereby.
The Salemme method for the preparation of porous polycarbonate
alld other resin membranes comprises the steps of:
(a) preparing a casting solution at room temperature consisting
10 of polycarbonate resin material and a casting solvent compo~ed of one or
more good solvents, the casting solution being stable at room temperature;
~ b) casting a layer of the casting solution so formed on a smooth,
clean surface or suppor t;
(c) permitting desolvation to occur for a predetermined time
15 interval from said layer;
(d) immersing said layer and support in a quenching bath liquid,
the quenching bath liquid being capable of dissolving the casting solvent
and causing swelling of the polycarbonate resin content of the layer while
being a non-solvent for the polycarbonate r~sin, the immersion step
20 initiating formation of a microporous membrane by entry of the quenchinD
bath liquid into said layer and exit of casting sol~ent theref~om;
(e) removing the microporous membrane from the quenchirlg bath;and
(f) removing the remaining casting solvent and quenching bath
liquid from the microporous membrane.
!07
The microporous films produced by the Exampl~s are said to be at
least as effective for filtration as those produced in accordance with the prior
art method of casting and maintaining in controlled atmosphere for extended
periods. Generally, the films are said to exhibit better flow rates and to be
5 more readily wettable than the prior art films
The response of these microporous films is measured in terms of
the foam-all-over point, which is the pressure required to cause foam to
develop over the surface of the film This method is commonly employed
in this art, and i8 referred to as the Bubble Point. Moreover, the process
10 for manufacture of these membranes is not susceptible of adaptation for
contlnuous production.
~ number of alcohol-insoluble polyamide resin membrane sheets
have been described, but to our knowledge none has been marketed. Where
sufficient information has been provided to permit duplication of the pro-
15 duction of these membranes, they have all been heavily skinned Membranesof alcohol-soluble polyamides have been made which are skinless, but they
have to be used with media which do not contain alcohol or a number of other
solvents in which they are soluble. Further, such membranes are not
capable of use after steam sterilization, a highly desirable attribute for
20 media used in large part for producing bacterially sterile filtrates. Hollow
fiber membranes made of polyamide resin are marketed in commercially
available equipment, but these are heavily skinned, and serve to accomplish
partial separations in the reverse osmosis range.
LovelletalU S. patentNo 2,783,894, patented~arch5, 1957, and
25 PaineU.S patentNo. 3,408,315,patentedOctober29,1968,provideaprocess
07
for producing alcohol-soluble polyamide membrane sheets using Nylon 4,
poly-~-butyrolactam. The term " alcohol-soluble" is used by these
patentees to refer to polyamide resins soluble in lower aliphatic alcohols
~uch as methanol or etha~ol, and is so us~ed in the présent specification
5 and claims. A solution of r~rlon can be cast as a liquid film and then con-
verted to a solid film which presents a microporous structure when dried.
An alcohol-water solution containing nylon is prepared and adjusted to the
point of incipient precipitation. The solution is brought to the point of in-
cipient precipitation by adding to the solution a solvent-miscible nonsolvent
10 which decreases the solubility of the nylon. This point is indicated when
a small amount of nonsolvent added to a sample of the solution causes an
obvious precipita~ion of nylon.
The nylon solution, adjusted to the point of incipient precipitation
and containing the proper additlves, is cast as a liquid film on an opticaliy
- 15 smooth surface of a solid base and then converted to a solid film by eg-
posure to an atmosphere containing a constantly maintained concentration
of e~cha~geable nonsolYent vapors, that is, vapors of a liquid in which
nylon is not soluble but which are exchallgea~le with vapors of the solvent
for the nylo~. The resulting membranes are7 of course, soluble in
20 alcohol, as well as in a considerable number of other solvents, and may
not be steam sterili~ed, which limits the scope of their usefulness.
HiratsukaandHoriguchiU.S. patentNo. 3,746,668, patented
~uly 17, 1973, also prepares membranes from alcohol solutions of poly-
amides which are alcohol-soluble, gelling the solution by addition of a
25 cyclic ether as a gelling agent, and drying the film. Alcohol-soluble
rQlatively low molecular weight copolymers of Nylon 6 and Nylon 66, and of
Nylon 67 Nylon 66 and Nylon 610 are used.
Marinaccio a~d Knight7 U. S. patent No. 3, 876, 738, patented April 8,
1975, describes a process for producing microporous membrane sheets from
07
a~;ohol-soluble and alcohol-insoluble polyamicles such as Nylon 6, poly-~-capro-
lactam, and Nylon 610, polyhexamethylene sebacamide, by casting a solution of
the polymer on a substrate and then precipitating the membrane, both steps being
carried out sequentially a concurrently in a ~uenching bath of nonsolvent liquid.
The nylon solution after formation is diluted with a nonsolvent for
nylon, and the nonsolvent employed is miscible with the nylon solution.
Marinaccio et al discuss polymer molecule aggregation in solution, and assert
that "the tightest or most nonporous polymer film is produced from a
solution in which there is no aggregate formation. "
According to Marinaccio et al, " . . . the resulting film strength is
primarily determined by the polymer concentration because of the larger
number of chain entanglements occurring at higher polymer levels. In
addition, for film cast from the ideal solution the "pore size" would increase
slightly with polymer concentration because of the increasing aggregation
15 tendency at higher concentrations. Aggregation in solution re~sults in film
porosity since the film as cast can be thought to consist of interacting
aggregated spherical particles. The larger the spheres, the larger the
voids in the film. Structurally $his is much like a box o~ tennis balls or
other nonspherical geomeb~ics fused at their point of contact. "
20 - As a first step, then, ~arinaccio et al control film porosity by
"control of the aggregation tendency in the casting solution. This is accom-
plished. . .by the addition of nonsolvent or other additives to change the
solvent power of the solution,hence influencing and controlling the aggregat~on
tendency of the polymer molecules. The interaction of these aggregates in
25 determining the resulting film structure is further influenced by the various
process variables previously maintained."
g
~l~L6~l~0~
'rhis is Marinaccio et al's theory, but it is not adequate to explain
what actually occurs, and is in many respects not consistent without actual
observations. Moreover, it differs from other more generally accepted
theories advanced to explain polyrner membrane formation7 as for instance,
Synthetic Polymeric Membranes, Kesting (McGraw Hill 1971) pp 117 to 15~.
Kesting's theory is more credible for a number of reasons; for e~ample, it
accounts for the very high voids volume of the rnembranes, which
Marinaccio's "tennis ball" theory fails to do;~ further it explains why only
relatively polar polymers are susceptible to membrane formation, which
10 again Marinaccio does not.
Marinaccio et al then-assert: "The selection of a solvent for a
selected film-forming polymer can be made on the basis of the foregoing
information. Determination of optimum solvent systems as well as other
process variables can then be made- on the basis of routine laboratory
15 experimentation. " However, dilution of the solution by addition of a
nonsolvent has a limit: "dilution with nonsolvent can be effected up to the
point of incipient precipitation of the nylon, but not beyond. " The casting
solutions are stable enough to be subjected to ageing periods of as much as
five to eight days, and indefinitely in some cases, but not so long that the
20 dissolved nylon separates.
The quenching bath may or may not be comprised of the same non-
solvent selected for preparation of the nylon solution, and may also contain
"slinall amounts'of the solvent employed in the nylon solution. However, the
ratio of solvent to nonsolvent is lower in the quenching bath than in the polymer
25 solution, in order that the desired result be o~tained. The quenching bath
may also include other nonsolvents , e. g., water. In all of the Examples,
the solvent utilized for the solutions is formic acid, but none of the quench
baths contained even a small amount of formic acid.
307
The Marinaccio et al process is said to di-ffer from conventional
methods of preparing microporous films in using more simplified castin~
solutions, but more impor-tantly in eliminating the slow equilibration step o-f
gelling in a high humidity atmosphere. In conventional processes this is a
critical step in the formation of the desired film structure. In the Marinaccio
et al process the film is cast directly in the quench bath, and immediately
quenched. By controlling the casting solution formulation as discussed abo~e
and controlling the quench bath variables including composition and temperature~film structure is said to be controlled. This technique forms the film structure"catastrophically"and is in direct contrast to the slow equilibrium technique
needed in conventional proce~ses.
In some cases Marinaccio et al suggest it may be desirable to pass
the cast film through a short air evaporation zone prior to the quench bath.
The technique could be used in those cases in which a graded cross-sectional
structure is desired in the film.
The product o~ Marinaccio et al has not been commercialized, and is
unavailable. The formation of a polymer film by direct immersion of the
casting resin into a quench bath is dlfficult, and it has not been economically
feasible to attempt to duplicate the Marinaccio et al process so that the
characteristics of the product couldbe studied, since such a study would
require constructing a rather elaborate apparatus. It is also nd:eworthy that
none of Marinaccio et al's Examples include formation of the film in a quench
bath, but instead are manually cast in individual laboratory tests onto glass
plates.
Tests were run using the glass plate method described by Marinaccio
et al, with delay periods between drawing the film and immersion in the bath
~ 6~ 7
varied from less than three seconds to as long as one minute; there was no
significant di-Eerence in product characteristics. It may therefore be assumed
that the film resulting from casting under the bath surface (representing
e2~trapolation to zero time) will not be different. With this in mlnd, the casting
5 resins of his ~xamples were formed as thin films, and with minimum delay,
always under one minute, so as to allow no significant loss of solvent by
evaporation, immersed into the baths described; in all cases the films
obtained ~vere heavily skinned.
A number of polyamide resin membranes have been used for reverse
10 osmosis and ultrafiltration, but all have pore sizes below 0. 1 ,u, and therefore
provide flow rates below the range useful in particulate and bacteria filtration.
Although the pores are small enough to remove microorganisms, such as
bacteria, such membranes are not used for this purpose, l~ut instead accom~
plish sucb tasl~s as reverse osmosis and ultrafillration, which are not
15 quantitative, and which can tolerate the imperfections which characterize
skinned nylon membranes.
Steigelmann and Hughes, U.S. patentNo. 3,9809605 patented
September 14, 1976, provides semipermeable membranes made-from mi~tures
of polyamides, especially N-alkoxyalkyl polyamides, and water-soluble polyvinyl
20 alcohols. The membranes are preEerably fvrmed as hollow fibers. The
membranes can be made from compositions containing the polymer components
and a di (lower alkyl) sulfoxide, e. g. ,dimethyl sulfoxide. The membranes may
contain complex-forming metal components. The membranes are useful for
separating chemicals from their mi}ctures by techniques using an aqueous
25 liquid barrier and cornple~-forming metals, e. g. ,for the separation of ethylen-
ically unsaturated hydrocarbons such as ethylene from close-boiling hydro
12
carbons, but such membranes have pore sizes too small to provide flow
rates useful in particulate and bacteria filtration.
It is an unfortunate fact that most available membrane sheets are
hydrophobic, i. e. 7 not readily wetted by water. Synthetic resin membrane
5 sheet has almost invariably been made of hydrophobic synthetic resin, and
retains the hydrophobic characteristic of the polymer from which it has been
made. The cellulose ester membranes ar e also hydrophobic. Of the available
membrane sheets useful in the particle separation range only ~egenerated
.
cellulose sheet and alcohol-soluble polyamide membrane sheet are hydrophîlic,
10 i. e., wettable by ~ater.
Brooks, Gaefke and Guilbault, U.S. patentNo. 3,9~1,810, proposed
a way around this problem, by preparing ultrafil~ation membranes made from
segmented polymers having distinct hydrophilic portions and hydrophobic
portions. Brooks et al suggest that if the casting solvent be a bett~r solvent for
15 the hydrophilic polymer segments than for the hydrophobic segments, the
resulting film or membrane will display a gross morphology in which the
hydrophilic portion of the system exists as a continuous phase while the
hydrophobic portion is present as a disperse phase. The membrane system
will include segregated domains of hydrophobic segments dispersed in a back~
20 ground of the hydrophilic polymer segments. By the ~ame token, if a casting
soluticnis selected such that it is a hetter solvent for the hydrophobic polymer
segments than for the hydrophilic segments, the phase relationships in resulting
films will be reversed and the film will not function as a membrane for a~ueous
media but will behave more as a hydrophobic film displaying virtually no
25 water permeabilityO
13
07
However, this expedient merely utilizes combinations of hydrophilic
and hydrophobic groups to achie~e water permeability~ and does not suggest
a way of modifying normally hydrophobic groups to improve water permea~
bility of hydrophobic polymers. Polyamides are not referred to by Brooks
5 et al as acceptable membrane materials -Eor their in~rention.
Yamarichietal, U.S. patentNo. 4,073,~33, describeahydrophilic
polyvinyl alcohol hollow fiber membrane with a relatively uniform pore size
distribution in the range from 0. 02 to 2 nnicrons, but these pores are not
interconnected, and the product serves for separation in the dialysis (high
10 molecular weight dissolved compound) range, rather than as a particle or
bacterial filter. - -
Since the bulk of filter applications for membrane sheet is in thefiltration-of aqueous media, it is essential to obtain an adequate wetting of the
sheet to facilitate filtration, but this is not easy to accomplish. Surface active
15 agents can be added to the medium being filtered, to enable the medium to wet
the sheet sufficiently to penetrate it for filtration. However, the addition of
foreign materials such as surface-active agents is not posei~le or desirable
in many applications, as for example, in assaying bacteria, since some
bacteria are killed by surface-active agents. In other applications, filtering
20 media cannot be adulterated by the addition of surface-active agent~ without
deleterious consequences.
Membrane sheets made of cellulose esters, which currently account
for over 95~c of all of the membrane sheet material sold,are inheren~y not
water-wettable; hence ~urface-active agents are added for water service.
25 Further, these membranes tend to be brittle7 and to counteract this, glycerine
14
~L6()~07
is added as a plasticizer, but this is also undesirable, since it will leach
into aqueous fluids, and poses a contamination problem which is unaccept-
able in ma~y uses.
In accordance wLth the invention, alcohol-insoluble polyamide resin
5 membrane sheet is provided that is inherently hydrophilic. This is a most
remarka~le properl;y, inasmuch as the alcohol-insolu~Le polyamide resin
from which the sheet is made is hydrophobic. The phenomenon occurs only
with a1cohol-insoluble polyamide resins having a ratio CH2: NHCO oP
methylene C~I2 to amide NHCO groups within the r~ge from about 5: ~ to
10 about 7:1. The reason why such polyamide resi~ membra~e sheet prepared
in accordance with the process of the invention is hydrophilic is not at
present understood, but it appears to be due to a spatial orientation of the
hydrophilic groups of the polymer chain that is fi~ed in the solid polymer
membrane surface as a result of the precipitation process. It may be related
15 to crystal structure or to solid structure, or to some spatial form of the
NH and/or CO groups on the surface of the membrane sheet, facUitating its
being wetted by water. The fact is that a drop of water placed on a dry
polyamide resin membrane sheet of the invention will penetrate into the sheet
and disappear within a few seconds. ~ sheet of the dry membrane place on
20 ~e surface of a body o~ water will be wetted through and may even sir~{ in
the water within a few seconds. If the membralle is complet~ly immersed
in water, 1~e membralle is wetted through in less than a ~econd.
The capabUity of a membrane's or subs$ra~e's being wetted by water
is detexmined by plaLcing a drop of water on the membra~e or substrate sur-
25 face. The angle of contact pro~rides a quantitative measure of wetting. A veryhigh angle of contact indicates poor wetting, while a zero angle of contact
defines complete or perfect wetting. The polyamide resin from which the
membranes of this in~ention are made ha~7e a high angle of contact, andL
are not wetted by water.
The wettability of these membranes is not a function of retained water.
Membrane specimens dried at 350F for 72 hours in an inert atmosphere, in
vacuum, and in air, are unchanged with respect to wettabili-ty by water. I-f,
however,they are heated to a temperature just below the softening temperature
5 of the membrarle ~to heat at a higher temperature would of course destroy the
membrane, since it would melt), the membrane reverts to a hydrophobic
material, which is no longer wetted by water. This suggests that the hydro-
philicity is a function of solid structure, and i~ obtained by the process of
membrane formation, probably during precipitation of the membrane in the
10 course of the process. It may be associated with crystal structure, or it may
only be associated with noncrystalline or amorphous solid structure, but it
does appear to be related to a physical orientation of the hydrophilic groups in
the polyamide chain, which orientation is losl; when the memhrane film is heated
to a high enough temp~rature to permit reorientation to a normal configuration,
15 in which $he material is hydrophobic.
It follows, of course, that during processing and drying it is importa~
not to heat the ~embrane above this temperature.
A further important characteristic of the polyamide resin membrane
sheets of the invention is their high fle~ibility. In the normal thickness range
20 in which they are useful, in the absence of an ext;reme state of dryness~ the~
can be folded back and for$h on themsel~es several times, without harm, and
without the addition of a plasticizer.
In the proce~ of the invention, polyamide re5in havlng a ratLo
C~2:N~ICO of methylene Cl-I2 to amide N~CO groups within the range from
~5 about 5:1 to a3~out q~ ; dissolved in a polyamide resill solvent, such as forr~ic
acid; a nonsolvent is ad~ed under controlled con~itions to achieve a nucleated
16
solution, and the resulting solution cast on a substrate in the form of a
film, and this film of solution is contacted and diluted with a liquid which is
a mixture of a solvent and a nonsolven~ for the polyamide resin. The
polyamide resin thereupon precipitates from the solution, forming a
5 skir~ess hydrophilic membrane sheet on the sub~trate, and the sheet can
then be wa~hed to remove the nonsolvent. The membrane can be stripped
off of the substrate and dried, or i~ the substrate is porous7 it ca~ be in-
corporated in the membrane or attached to the membralle to serve as a
permanent support, and is then dried with the membra~e.
The conditions under which the polyamide resin is precipitated
de$ermine the skinless nature of the membrane, as well as its physical
characteristics, i.e., the size, length alld shape of ~he through pores of
the membrane. Under certain conditions, a membrane is formed which
has 1;hrough pores extending from surface to surface tha~ are subst~l~ially
15 uniform in shape and size. Under other oonditions, the through pores are
tapered, being wider at one surface alld narrowing towards the other
surface of the membrane.
Under conditions outside the scope of the ir~ve~tion, still another
form of the membrane is obtained, having a dense skin penetrated by pores
2~ o~ smaller diameter $han the pores in the remainder of the sheet. This
skin is normally on one side of the membrane sheet~ but it can be on both
sides of the mem~rane sheet. Such skinned membranes are con~rentioTlal in
the art, e~hibit relatiYely higher pressure drop and other poor filgration
characteristics, and a~e undesirable.
Thus, by colltrol of the method by which the casting resin is
nucleated, and of the precipitation c~nditions, it is possible to obtain
17
~ 7
hydrophilic polyamide resin membranes with through pores of desired
characteristics, either uniform from face to face, or tapered, with
larger pores on one face transNioning to finer pores on the other face.
The formation of a polyamide memb~ane ha~ing uni~orm pores
5 or tapered pores without a skin on either surface is also ~emarkable. As
shown by l~ie Michaels pa$ent No. 3, 615, 024, and Marinaccio et al
No. 3,876,738, precipitation of a polyamide resin membrane in a non~
solvent is known to result in a skinned membrane. The formation of a
hydrophilic s~ ess poly~nide resin membrane by this process has not
10 previousl~ been achieved.
The process of the inve~tion for preparing from hydrophobic
polyamide resin a skir~ess microporous polyamide membrane having
absolute particle removal ratings of 0.10 ,~M to 5 ,u~ or larger in a solid
form that is hydrophilic and remains hydrophilic until heated to a tempera-
15 ture just below its softening point comprises preparing a solution in apoly~n~ide sol~Tent of an alcohol-insoluble poly~mide resin having a ratio
CH2:NHCO of methylene CH2 to amide NHCO groups within the range from
abou$ 5:1 to about 7:1, inducing nucleation by dilution of the solution with
a nonsol~ent liquid under controlled 0nditl0ns o~ solvent and nonsolvent
20 and resin concen$ratioII9 temperature, mixing intensi~y, addition time
and system geometry such that a visible precipitate of polyamide resin
forms during the addition of the nonsolvent, with or without visibly
complete redissolutlon of the precipitated polyamide resin; removing any
undissolved resin by filtration; spreading the resulting solution on a
25 substrate to form a thin film thereof on the substrate; con~acting the film
wi~ a mi~ture of nonsolvent liquid containing a substantial proportion of
18
07
solvent for the polyamide resing thereby precipitating polyamide resin in
ffle form of a thin skir~ess hydrophUic membrane; and washing and drying
the resulting membrane.
In a preferred embodiment of this process, the solvent for the
5 polyamide resin solution is formic acid and the nonsolYeDt is water, aIld
the polyamide resin solution film is contacted with the nonsolvent by
immersing the film carried on the substrate in a bath of nonsolvent
comprising water containin~ a substantial proportion of formic acid.
The invention in another preferred embodimen~ provides a process
10 for preparing ski~ess hydrophUic alcohol-insoluble polyamide membrane
sheets having pores that are substantially uniform from surface to surface,
which comprises preparing a solution in a polyamide solvent of an alcohol-
insoluble polyamide resin ha~ing a ratio of CH2: NHCO of methylene ~H2 l:o
amide N~ICO groups within the range from about 5:1 to about 7:1; inducing
15 ~cleation by dilution of ~e solution while controllin~ sol~ent and nonsolvent
and resin concentration, temperature, mixing intensity, addition time and
system geometry to obtain a ~Tisible precipitate of polyamide resin during
the addition of the diluent,with or without visually complete redissolution
of the precipitated polyamide resin, thereby forming a casting solution;
20 remo ~ing any undissolved resin by filtration, spreading the castin~
solution on a substrate which is nonporous and whose surface is wetted ~y
~e casting solution and preferably also by th~ nonsolvent-solve~ mi~ture
to form a thin film ~hereof on the substrate, contacting the film with a
mixture of nonsol~ent liquid containing a subs~antial proportion o~ solvent
25 for ~e polyamide resin, thereby precipitating polyamide resin in the form
of a thin skinless ~rdrophilic membrane, a~l washing and drying the
resulting membrane.
19
~i~L~ 307
Further, a continuous process is provided for preparing skinless
hydrophilic alcohol-insolu~le polyamide membrane sheets which comprises
preparing a solution in a polyamide solvent of an alcohol-insoluble poly-
amide resin having a ratio CH2: N~ICO of methylene CH2 to amide NHCO
5 groups within the range from about 5:1 to about 7:1; inducing ~cleation
by dilution of the solution with a nonsolvent while controlling solvent and
nonsol~ent and resin concentration, temperature, mi~{ing intensity,
addition time and system geomQtry to obtain precipitation of polyamide
resin ~uring the addi$ion of the nonsolvent, with or without visually com-
10 plete l edissolution of the precipitated polyamide resin, ~ereby forminga casting solution; removing any undissolved resin by filtration; spread-
ing the casting solution on a substrate which is nonporous and whose
surface is wetted by the casting solution and preferably also by the
nonsol~Tent-solvent mixture to form a thin film thereof on the substrate;
15 contacting the film with a bath of nonsolvent liquid containing a substantial
proportion of solven~ for the polyamide resin, thereby precipitating
polyamide resin in the form of a thin skinless hydrophilic membrane;
aIld continuously washing and drying the resulting membr~ne, while
maintaining constant the rel~tive proportion of solvent and nonsolverlt
20 liquid in the bath. In a preferred embodiment, the rates o withdrawal and
addition of sol~ent and nonsolvent to and from the bath are maintained
substantially const~nt.
The invention further provides a process for preparing skinless
hydrophUic alcohol-insoluble polyamide membrane sheets having multi-
25 membrane layers, which co~nprises preparîng at least two startingsolutions in a polya:mide solvent of alcohol-insoluhle polyamide resin
0~l
- having a ratio CH2:NHCO of methyleIle CH2 to amide NHCO groups within
the range from about 5:1 to about ~:1; inducing nucleation by dilution of
the solutions wLth a nonsolven$ while controlling solvent and nonsolvent
and resin concentration, temperature, mi~ing intensity, addition time and
5 system geometry to obtain a visible precipitate of polyamide resin d~ring
the addition of the nonsolvent, with or without visibly complete redissolution
of the precipitated polyamide resin; removing any undissolved resin by
filtration; spreading the resulting solution on a substrate which is non-
porous and whose surface is wetted by the casting solution and prefera~ly
10 also by the nonsol~rent-solvent mixture to form a thin film thereof on the
substrate; contacting the film with a migture of nonsolvent liquid containing
a substantial proportion of solvent for the polyamide resin, thereby pre-
cipitating pol~amide resin in the form of a ~hin ski~ess hydrophilie mem-
bran~; washing the resulting two membranes; combining the t~o mem-
15 branes so ~ormed as a dual layer; and dryillg the dual layer underconditions of restrain to prevent more than minor reduction of the length
and wid~:h of the membrane; the membranes so dried forming a single
sheet with particle removal characteristics æuperior to those of the
Individual layers.
The membranes thus attached can ha~re the sar~e or difering
porosities, and the m~mbrane layers can be selected from membranes
having tapered pores and mem~ranes having uniform pores, in alYy
combination, supported or unsupportedO
The two combined mem~ranes can be obtained from a single
25 roll of filter medium, and when combined with matching faces in con-
tact form a sheet which is symmetrical, and which provides equal
filtration chaxacteristics regardless of which face is upstream.
21
07
The Invention also provides several types of polyamide resin
membrane products. One prefelred embodiment is a hydrophilic micro-
porous polyamide membrane comprising a normally hydrophobic polyamide
resin having a ratio CH2:NHCO of methylene CH~ to amide NHCO groups
5 within the rangefrom about 5: 1 to about 7:1 in a solid structure that is
hydrophilic, having absolute removal ratings wi$hin the range from about
0. 1 uM to about 5 ,uM, and a thicl~ness wi thin the range from about
0. 025 mm to about 0. 8 mm.
These hydrophilic microporous polyamide resin membranes can
10 have pores e~tending from surface to surface in a relati~ely uniform
structure, or in a tapered pore structure.
Also provided are hydrophilic polyamide resîn membranes that
are supported by the substrate on which the polyamide resin membrane is
formed, either imbedded therein, or having the substrate attached to one
15 face thereof.
In addition, the invention provides microporous polyamide resin
membrane composites having a plurality of polyamide resin membrane
layers, formed of meml~ranes preparad separ~tely ~y precipitation on
separate substrates and then bonded together by drying two or more
20 layers main$ained in close contact.
In all of these em~odiments, $he polyamide resins having a ratio
C~2:NHCO of methylene CH2 to amide NHCO groups wifflin the range from
a~out 5:1 to about q: 1 include polyhe~amethylene adipamide ~ylon 66),
poly-E~caprolactam (Nylon 6)~ polyhe~amethylene sebacamide ~ylon 610),
25 poly-7-aminoheptarloamide (Nylon 7), polyhe~amethylene a~eleamide
22
07
(Nylon 69), and mi~tures of two or more thereof, as well as mixtures
thereof with higher polyamide homologues such as polyhe~methylene
sebacamide ~Nylon 612) in proportion such that the mixhlre has an
average of CH2:NHCO ratio within the stated range. The first three
5 polyamides, Nylon 66, Nylon 6 and Nylon 610, are preferred.
Another purpose of the invention is to pro~Tide a procedure for
quantitati~Te characterization of uniform pore membranes for their
abilit~ to provide sterile ef~luent when challenged by a stated number of
a gLven microorganism. This procedure is applicable to uniform pore
10 distribution membranes made of other than polyamide resins, and using
other processes.
In the drawings:
Figure 1 is a graph showing in a qualitative manner the relation-
ship between the degree of nucleation of the casting resin ~olution alld
15 the pore diameter of the resulting membrane;
Figure 2 is a graph showing the relationship for a uniform pore
membrane between titer reduction, defined as ~he ratio of Pseudomonas
diminutiae bacteria contained in the inflNent liquid to ~at obtained in the
effluent liquid, a~d the number of layers of uniform pore fUter medium
20 through which the bacteria laden liquid is passed.
Figure 3 is a graph showing ~e relationship obtained when a
we~ted membrane is pressurizedby agas, and the ratio airaiprrflsure
is plotted against air pressure applied. The qualltil~T KL is defined by
~e broken line of F_gure 3 .
23
~:~6~0a7
Fi~ure 4 is a graph showing the relation between T.~ and KL, where
T
(or, Log TR ~ ~ Log TR)
10 where
t is the thickness, in thousands of an inch, of the uniform pore
membrane which shows a titer reduction, as ~defined above, equal to TR;and
TR is the calculated titer reduction for a 0. 001 inch thick membrane
of equal pore size~
KL is the pxessure, measured inpsi, atwhich air flow through the
water wet membrane increases very sharply (see Fi~ure 3); and
K is the ~ralue of K corrected to correspond to that of a
L5 L
membrane 0. 005 inch thick, using the empirically determined correction
factors listed in Table 1.
TABLE I
:
Measured Thickness .
inch Corr ection ~actor
0.002 1.10
. 003 . -- 1. 044
. 00~ 1. 019
O. 005 ~L. 000
0.006 0.985
0. 008 0. ~62
O. ûlO 0. 9~6
0.015 0.920
24
- ~160~07
The curve of Fi~-~lre 4 represents the results of measuring the KL
and TR for forty-five different specimens made by the process of this invention.
Figure 5 is a scanning electron rnicrograph (SEM) at 1500~magnifi-
- cation of a membrane with uniform pores, made by the process of this invention,
having a KL of 47 psi, t = 0. 003~ inch, and an estimated TR of 3 x 10l8 for
Pseudomonas diminutiae organism The center portion of this micrograph
shows a section through the thickness of the membrane, in-which the pore
sizes are seen to be uniform from surface $o surface. The upper and lower
micrographs show the upper and lower surfaces respectively adjacent to
10 the section, and the pore size at each of these surfaces may also be seen
to be equal.
Figure 6 is a scanning electron micrograph at lOOQXmagnification
of another membrane with unlform pores made by the process of this in-
vention, having a KL of 40 psi, t = 5. 6 mils, and an estimated TR 8 ~ 5
15 for Pseudomonas diminutiae organism. Similar to Figure 5, the center
portion is a section through the membrane sh~wing uniform pore size from
surface to surface, and the upper and lower views show t~e adjacent upper
and lower surfaces, again indistinguishable with respect to pore size.
Figure ~ is a scanning electron micrograph at lOOOXmagnification
20 of a membrane with tapered pores made by the process of this invention.
This membrane is 81 ,uM (0. 0032 inch) thick, and the upper portion of the
section may be seen in the central part of the SE~ to be considerably
smaller in pore diameter than the adjacent material, with the pore diameter
gradually transitioning to the larger size. Comparing the top and bottom
25 vie~s, the pore diameters in the upper surface are substantially smal~er
than those on the lower surface.
~L~6C~07
- Figure 8 is a scanning electron micrograph at 1500 X magnifi-
cation of a lightly skinned membrane, of the l~pe obtained when baths
outside of the range of this in~7ention are employed.
Figure 9 is a similar microgra~?h of a ~ re heavily skinned
5 mem~rane.
Figure 10 is a graphical representation of the relationship
between
(a) KL ~ a particle removal rating parameter of membranes
made by the process of this invention, defined by this invention;
~3) the mixing intensity7 expressed as revolution per minute
(rpm) of the in-line mixer used to process a 15. 5 Yc solution
of resin in 98. 5~c formic acid to obtain the casting solution
used to produce the membra~es; and
(c) ~e formic acid concentration of t~e resulting casting
solution.
~ 1i1hile the various polyamide resins are all copolymeræ of a
diamine and a dicarbogylic acid7 or homopolymers of a lactam of an
amino acid, they vary widely in crgstallinit~ or solid structure, melting
point, and other physical properties. It has been determined in accordance
20 with the invention that the process of the invention is applicai)le oT~y to
polyamides h~ving a ratio CH2:NHCO oX methylerle CH2 to amide N~ICC
groups within the raxlge from about 5:1 to a~out 7:1 exemplified by the
polyamides enumerated pre~riously. The preferred members of this
group, copolymers of he~amethylene diamine and adipic acid (~ylon 66),
25 to copolymers of he~nethyl0ne diamine alld sebacic acid ~Nylon 610),
a~d to homopolymers of poly-~-c~prolactam ~Nylon 6), readily produce
~6
~L~6(~(~0'7
skir~ess hydrophilic alcohol-insoluble polyamide resin membra~es. For
reasons which are not understood, all members of this limited class of
polyamide resins are quite susceptibl0 to precipitation under the process
conditions of the invention to form hydrophilic membrane sheets.
These polymers are a~ailable in a wide variety o~ grades, which
vary apprecia~ly with respect to molecular weight9 within the range from
about 15, 000 to about 42, 000, and in other characteristics. The formation
of a hydrophilic membrane appears to be a function not of these character~
istics, but of the chemical composition of the polymer, i. e., the spacing
10 and arrangement of the units composing the polymer chain. The especially
pre~erred species of the units composing the polymer chain is polyhexa-
me~ylene adipamide, and molecular ~eights in the range above about
30, 000 are preferred. Polymers free of additires are gen~rally pre-
ferred, but the addition of antioxidants or simUar additives may have
15 benefi$ under some conditions; for e2~3mple, addition of the a~tioxidant
E~yl 330 (1,3,5-trimethyl-2,4J6-tris [3~5-di-tert-butyl-4-~ydrox~
benzyl~ ben~ene) has been shown to e~tend the life of polyamide mem-
~ranes exposed to extreme ogidative hydrolytic conditions.
The polyamide resin solution fr~m which the polyamide membrane
20 film is precipita~ed can be a solution in any sol~er~ for the polymer~ These
solvents are well ~own, and are themselves no part of the instant in~ention.
~ preferred solven~ is formic acid at any temperature from its freezing
point to its boiling pomt. O~er suitable solvents are: other liquid
aliphatic acids such as acetic acid and propionic acid, and halogenated
25 aliphatic acids such as trichloroacetic, trichloropropionic, chloroacetic
acid, dichloroacetic acids, phenols such as phenol, the cresols, and their
2~
07
halogenated derivatives; inorganic acids such as hydrochloric, sulfuric,
hydrofluoric and phosphoric; saturated aqueous or alcohol solutions of
alcohol-soluble salts such as calcium chlorid~, magnesium chloride and
lithium chloride; hydro~ylic solvents including halogenated alcohols
5 (trichloroethanol, trifluoroethan~l), benzyl alcohol, and polyhydric
alcohols such as ethylene glycol, propylene glycol7 and glycerol; and
polar aprotic solvents such as ethylene carbon~tea diethyl succinate,
dimethyl sulEoxide and dimethyl formamide.
The polyamide resin solution, hereafter refereIlced as the
10 starting resin solution, is prepared by dissolution of the polyamide resin
to ~e used in the membrane in the desired solYent. The resin can be
dissolved in the solvent at ambient $emperature, but a higher temperature
may be used to accelerate dissolution.
If the starting resin solution is to be stored for more than a few
15 hours, water in eEcess of a~out 1 to 2~c should not be present, as other-
wise a slow hydrolysis of the polyamide resin takes place, resulting in
a~ undesirable reduction în molecular weight of the polyamide. In general,
-the amount of water in this event should be less than 2~c, and preferably
the solution is water-free. If water or formic acid-water mixture is
20 added to accomplish nucleation, it can be added just prior to casting,
preferably within a~out five to six~y minutes of the casting oper~tion.
The casting resin solution is prepared from the starting resin solution
by diluting 'It wiffi a nonsolvent, or with a mixture of solvent and nonsolvent.
28
~6~(~()7
The state of nucleation of the resulting casting resin solution is strongly
affected by the following factors: -
1) Concen~ation, temperature and molecular weight of the starting
resin solution;
2) Composition and temperature of the nonsolvent, or of the
nonsolvent-solvent mixture.
3) The ràte at which the nonsolvent, or nonsolYent-solvent mixture,
is added.
4) The intensity of mixing during the addition.
5) The geometry of the appaxatus in which the mixing is accomplished.
6) The temperature of the resulting casting resin solution.
The casting resin solution so prepared is then formed into a thin film by
casting it onto an appropriate substrate, and the film is immersed with
minimum delay into a bath containing a nonsolvent for the polyamide resin,
15 together with a substantial proportion of solvent for the resin. Tf ths nonsolvent
in the bath is water, and i the solvent is formic acid, the presence of at least
about 20~C and usually of at least 30 to 40~ of formic acid is desired to
prevent formation of a skinned membrane, which occurs at lower concentra-
tions of formic acid.
The stability of the casting ~esin solution varies greatly depending on
the method used to prepare it. For example, casting resin solution prepared
under small scale batch conditions tends to be relatively unstable; for example,the characteristics of the membranes it produces will be quite different if it is
cast as long as five to ten minutes after it has been prepared, or it may
transform to a noncastable semi-solid gel within 10 minutes or less. On
29
.6~)7
the other hand, casting resin solution prepared using a continuous in~line
mîxer, which can produce a membrane of equal characteristics, tends to be
stable for a period of an hour or more. Casting resin solutions prepared in
this way should, however, be used within an hour or less, particularly if
5 maintained at ele~rated temperature, to prevent substantial reduction in
molecular weight of the polyamide resin which will otherwise occur due to
the presence o water in the acid solutiong with resultant hydrolysis.
Eith~r of the above me$hods may be used to produce casting resin
solutions which function equally when cast as membranes~ and regardless of
10 which is used the addition of the nonsolvent is accompanied by the appearance
of a ~-isible polyamide resin precipitate, in order to produce a useul, properly
nucleated casting resin solution. Casting resin solutions prepared by o~her
means, for example, by dissolving the resin pellets in a solution of formic
acid and water, or by adding the nonsolvent in a manner such as not to
15 produce such a precipitate, do not produce useful membranes~
IJseful membranes are those with uniform or tapered pore structures,
skinless, with permeabilities to air and water such that substantial quantities
of fluids can be passed at low pressure differentials, while providinv a
required dearee of filtration. A convenient index of useful~ess may be
20 obtained by considering the permeabilities to air and to water of uniform pore
cellulose ester membranes now on the market made by the so-called dry
(evaporative) process. These are sho~Yn in Table ll under, to~ether with
typical permeabilities of similar range media made by the process of this
invention.
~5
37
TABLE TI
Typical Flow Rates of IJseful MembraneS
~ - Polyamide mem-
Absolute removal Flow per sq. ft. Commercial cellulo5e branes of this
rating, micrometers per psid ester membranes invention
- ~pm H2O 0.04 0.17
0.1 cfm air~ ~ ~.4 2.5
gpmHO 0.38 . 0.57
0. 2~ _ 2
cfm air 8. 0 8. 4
0. 45 cfm air 1 0 l j0
Membranes having significantl~ low~r flow capacities for equal
10 removal characteristics, when compared with currently marketed membranes,
are not widely commercially acceptable, and have been defined, for purposT?s
of this discussion, to be out of the useful range.
It is an important feature of this invention, that the conditions ar~ .
described for achieving a casting solution with controlled degree of nucleation
15 to make membranes with useful pressure drop characterîstics.
We use herein the terms "nucleation" and "~ta~e of n~lcleation" to
account f~ the discovery that
(a) casting resin solutions can be prepared with a wide variation of
composition with respect to resin, solvent, and nonsolvent concenl:ra-
tions, whic.h yield identical or nearly identical membranes; and
(b) casting resin solutions can be prepared~ which have equal resin?solvent and nonsolvent concentrations, which are then ~ast at equal
temperatures into the same bath, yet yield very different membranes;
in fact, the resulting membranes can run the gamut ~om "not usefulT'
in the sense of ha~ing very significantly lower flows-by factors of 2 to
5 or more~ compared with Table I, through $he range ~om 0.1 ~M
absolute or coarser, producing membrane5 in all those ranges with
good flow capacities, for example, equal to those listed on Table II.
~1 ~
~L~L6~ ;)7
Since the preparation of casting resin solutions capahle of producing
membranes with flo~v properties in ~he useEul range has been observed to
invariably be accompanied by the local precipitation and at least partial
redissolution of solid resin, and since it is well known to those familiar
with the chemical arts that the characteristics of a solid precipita~ed from
solution can be greatly influenced by the presence or absence of submicro-
scopic nuclei, we have chosen to use the tlerm "state of nucleation" to
distinguish casting solutions having equa1 composition, but diverse results,
as described in paragraph ~b) above, and to account as well for the observation
of paragraph (a).
The assumption that nucleation accounts for the differences in
behavior of membranes made from casting resin solutions of equal composition
is confirmed by the results of an experiment in which a stable casting resin
solution was prepared, with a degree of nucleation controlled to yield a 0. 4 ,u~
absolute membrane. A portion of the casting resin solution was subjected to
fine filtration to determine whether nucleation beha~rior would be affected, andthe properties o membranes cast from the two lots of casting resin solution
were compared.
- Examples 58 and 59 show the results of this experiment;
product characteristics are greatly altered by fins filtration; the finely
filtered casting resin solution produces a membrane with a ver~ poor ratio
or flow capacity to remo~al rating; $he ~p of sample No. 59 is more than
three times higher than that for a similar membrane made using a properly
nucleated casting resin solution o this invention.
This result supports the theory that resin nuclei are developed during
the controlled-condition dilution used to prepare the casting resin solution,
32
07
whose number, size, or other characteristics strongly influence the
characteristics of the membrane generated ~y that casting resin solution,
and that at least a po. L;on of these nuclei were removed by fine filtration.
It should, however, be understood that we have not unque~tionably
5 established that nucle~tion is the only explanation Eor the observed results,
and that they could be caused by phenomena other than nucleation.
The viscosity of the casting resin solution is preferably ad3usted to
between about 500 centipoises and 5000 centipoises at the temperature existing
at the time it is cast as a film~ ~iscositi~s below about 500 cp allo~v some of
1~ the cast film to float off as a liquid to the surface of the bath, when it forms a
filmy precipitate, thereby adversely af~ecting cast membrane propertie~s
and fouling the bath. Viscosities much above 5000 cp, For example,
100, 000 cp, are not Ileeded to obtain a smooth, coherent cast film7 but
are helpful in casting membranes where no substrate is used, for example,
15 hollow fibers, or unsupported film.
Solutions of a viscosity well above 5000 cp at the castLng temper~ture
can be cast without difficulty7 however, the preferred vissosity limit is about
5000 cp, since at higher viscosities the energy input to the mixture when a
nonsolvent is blended with the polyamide resin solution is very high, with the
20 result that the solution can l~each e2~cessively hîgh temperature, with ensuing
operating problems. Moreover, the pumping of the startingrpolyamide resin solution
to the casting operation becomes progressively more difficult, as viscosity
increases~ Also, manipulation of the casting resin solution within $he
reservoir from which the resin is cast as a film on the substrate becomes
.
33
troublesome, if the viscosity is very high. When a porous substrate is used,
with the intention of completely impregnating it with casting resin solu-tion,
viscosities much above about 3000 cp can cause improper penetration, and
the resulting product has undesirable voiAs.
The temperature of the casting resin solution is not sritical, and
useful membranes have been made over the range from about 85C downward.
Under some circumstances, some~vhat highler flow rates relative to removal
rating are obtained by reducing the resin temperature to a lower value prior
to casting the films.
After the cast film of li~uid enters the bath, a precipitation process
occurs, whose mechanism is not completely underslood. The nonsolvent
~nixture of the bath diffuses into the cast film, and the solvent mixture in
the casting resin solution diffuses out of the film into the bath, but it is notunderstood why this results in a uniform pore size throughout the thickness of
the film ~vhen the bath solvent-nonsolvent ratio is held within certain limitsO
If the bath contains only nonsolvent (such as water, alcohols or
organic esters), or nonsolvent with a small proportion of solvent ~e. g. water
with less than 15 to 20% of formic acid) precipitation occurs very rapi~ly,
and the solid membrane is formed within a few seconds, typically in less than
1 to 10 seconds. Membranes made in this manner are heavily skinned,
regardless of the mode of preparation of the casting resin solution, and are
undesirable~
If the bath contains about 43 to 55% oE formic acid in aqueous solution, and
the casting resin solution is properly nucleated as described herein, the resulting
membrane will be uniform in pore structure from face to face, provided only
34
that if cast on a solid substrate, the surface of that substrate be wetted by
the casting resin solution and by the bath solution. The time required ~or the
film to form under these circumstances is a function of the following:
(a) Casting resin solutions which produce membranes which ha~e high
KL values (e.g. in excess of 100 psi) set very rapidly, e.g. in less
than 10 seconds. Less highly nucleated casting resin solutions,
producing membranes with KL values of about 40 to 50 psi will typically
set in the 10 to 20 second range, and the setting time continues to
increase as KL decreases, such that membranes about 0. 006 inch
thick with KL ~ralues o under 20 psi require about 5 m;nutes or more
to set, and still lower KL's require still longer periods.
O Thickness of the cast film is an important parameter, setting
times being shorter for thin films.
(c) Use of lower casting resin solution temperatures results in aster
settingO
(d) Setting is faster at the low end of the 43 to 55% recommended
range, and can be further speeded by use of bath concentrations
of less than 43% formic acid, at the cost of only slight deviation
from pore uniformity.
As the bath concentration decreases to and below the 40 to 43%
range, the membranes become progressively more asymmetric, progressing
from uniform as shown in Figures 5 a~d 6, to tapered pore as sho~Tn in
Figure 7~ to skinned as shown in Figure 8? to heavily skinned as sh~wn in
Figure 9. Operation at formic acid concentrations much lower than those
5 proc~ucing tapered pores as exemplified by Figure 7 is undesirable.
The formation of the membrane from a casting resin solution can be carried
3~
~ )7
out as an intermittent or batch operation or as a continuous or semicontinuous
process. A small scale operation may be most conveniently carried out as a
batch operation, while at high production rates a continuou~ or semicont~nuous
operation is more convenient. In all types of processes7 it is important to
5 carefully control all of the operating parameters to ensure a uniform product,
including operating temperatures, and relative proportions of resin solution
and nonsolvent liquid. The control of conditions of non-solvent addition are
particularly important, includlng the geometry of the ~pparatus, the rates of
~:low, and duration and intensity of mixing; also the interval between nonsoluent
10 addition and casting of the resin film must be controlled. Such controls can
be established by trial and error e~perimentation without undue difficulty by
those skilled in this art, taking into account the following considerations:
It is important that the casting resin solution be clear, and free from
suspended material, before being spread upon the substrate to form a film.
15 If suspended material is present, such as undissolved resin particles, these
are removed by screening or filtration before casting.
Any type of subs~ate or support can be used as a surface on which
the casting resin solution is cast to foxm the solution film. If a nonsupported
membrane film is the desired product, then the substxate should have a
20 sur-~ace to ~vhich ~he membrane does not adhere, and from which the
36
07 )
membrane film can readily be stripped at the conclusion of the drying
operation. Strip~ability usually requires that the substrate surface be
smooth-surfaced9 and nonporous. When the solvent is one with a relatively
high surface tension, such as formic acid, and the nonsol~ent also has a
5 relatively hi,,h surface tension (as3 for example, water), it is important
that the nonporous surface on which the filrm is cast be wettable, i. e,
have zero or near zero allgle of contact, when contacted by the casting resin
solution, and preferably also by the bath as well. Failing this condition,
a skin will form on the membrane on the substrate side, with undesirable
10 effect on membrane properties~ Such temporary subs~ate or support
surfaces can be of a suitable ~naterial, such as glass, metal or ceramic.
Plastics, such as polyethylene, polypropylene, polyester, synthetic and
natural rubber, polytetrafluoroethylene, polyvinyl chloride7 and simUar
materials ~re not inherently suitable, as they are not wetted by the casting
15 resin and nonsolvent, but these can be rendered suitable by application of
an appropriate oxidative nr similar surface treatment. A corona discharge
can, for e~ample, be used to treat Mylar ~polyester~ film, and polypropylene.
The substrate can be made of or merely surfaced with such materials.
If the substrate is to form a part of the final membrane film, as a
20 permanent supporting layer, then it should be o~ porous material that preferably
is wetted by the casting resin solution, so that the casting resin solution will
penetrate it during the casting of the solution on the subs~ate, a~d become
firmly attached thereto during precipitation of the polyamide membrane film.
It is not essential howe~er that the substrate be wetted; if it is not wetted,
25 the polyamide resin film will be largely confined to the surEace of the supportJ
* Trademark 37
V07
~ut is nonetheless aclherent thereto. Such substrates ca~, ~or example, be
of nonwoven or woven fibrous material, such as nonwoven mats and bats,
and woven te~tiles and cloth, as well as netting of various types, including
extruded plastic filament netting, papers, and similar materials.
As permanent supports which are not wetted by the casting resin '6
solution, fine-pored nonwoven webs can be used, made from fibers with poor
wetting characteristics, such as, for example, polypropylene or polyethylene.
The resin solution îs cast as a film onto the nollwoven web, and since it does
not wet the fibers of the web, it is carried on its sur~ace. The substrate
carrying the casting resin solution film on its lower ~urface is plunged into a
bath of nonsol~rent liquid or allowed to float on the surface of the bath, and the
membrane film precipitated onto the substrate. The resulting film has
good adhesion to the substrate, and the substrate has very l~ttle or no effect
on the pressure drop for fluid ~low through the membrane.
lS In the case of permanent supports which are wetted by the casting resin
solution, the fibers of which the substrate is made should have a relatively
high critical surface tension, such that the casting resin sol7ltion film wili
~ompletely permea~e the suppor$ing web, and the resulting membrane
precipitates in arld around the fibrous material, and is permanently
supported thereb~, since the material of the support is embedded in the
membrane. The resulting membrane has a somewhat higher pressure drop
when tested with flowing fluid, but the increase compared with the unsupported
membrane is small, if the supporting web has an open structure.
Suitable wetted substrates that can serve as permanent supports for
the membrane include polyesters, as a nonwoven fibrous web or as a woven
web, using monofilament or multifilament yarn, the monofilaments being
38
17
preferable in terms OI open structure and lower pressure drop; also
polyimide fiber woven webs, woven and nonwoven webs of aromatic polyamides
or Nom~, and other relatively polar fibrous products such as cellulose,
regenerated cellulose, cellulose Psters, cellulose ethers, glass fiber, and
5 similar materials.
Cellulosic and synthetic fiber filter papers can be used, as well as
perforated plastic sheets, and open mesh expanded plastics such as Delnet
or similar extruded and thereafter e~pandecl nefflngs. If the substrate is
relatively coarse or in a very open w0ave s~ucture, even if the fibers are
10 not well wetted b~r the resin solution, the substrate may nonetheless be
embedded or embraced by the membrane material in the ~Einal supported
membrane productj such relatlvely poorly wetted materials as polypropylene
and polyethylene can function as embedded substrates if they have a
sufficiently open s~ucture. If a polyolefin substrate has a relatively smaller
15 pore size, for ~ample, about 30 microns, the casting resin solution wlll not
penetrate in$o it, but will instead form a membrane external to, but adhered
to, the polyolefin substrate.
In a continuous process, the substrate can be in the form of an
endless belt, which circulates through the entire film-forming operation, from
casting of the casting resin solution film into and through a pq~ecipitating bath
of the nonsolvent liquid, and then through the bath liquid removal step. A
corrosion resistant metal drum, or endless metal belt can be used, but the
surfaces on which the ~Eilm is cast should be treated or coated so as to make
them wet~able.
The nucleated casting resin solution can be cast or spread out upon
the substrate in the de9ired ~Eilm thickness using a conventional doctor blade
* Trademarks
39
~6~l~1)7
or roll, kissing or squeeze rolls or other co~entional devices, and then
contacted with the ~ath liquid with as little delay as possible.
The choice of nonsolvent liquid depends upon the sol~ent utilized.
The preferred nonsolvent for producing nucleation in the polyamide resin
5 solution is water or wa~er-formic acid mixtures. lIowever, any substance
is suitable which is soluble in water and reduces the sur~ace tension of
water. Other nonsolvents anclude formamides and acetamides, dimethyl
sulfoxide, and similar polar solvents, as weli as polyols such as glycerol,
glycols, polyglycols, and ethers and esters thereof, and mi}~tures of such
10 compounds. Salt~ can also be added.
~ ?ollowing precipitation, the membran0 film is washed to remove sol-
vent. Water ls suitable, ~ut any ~olatile liquid in which the solvent is soluble
and that can be removed during drying can be used as ~e washing liquid.
One or several washes or baths can be used as required to reduce
15 sol~ent content to below the desired minimum. In the continuous process,
w~ash liquid flow is countercurrently to the membrane, which can, for example~
be passed through a series of shallow washing liquid baths in the washing stage.
The amount of washing required depends upon the residual sol~rent
content desired in the mem~rane. If the solvent is an acid such as formlc
20 acid, residual formic acid can cause hydrolysis during st~rage of the poly~
amide of which the membrane is composed, wLth a conse~uent reduction in
molecular weight; therefore, the washing should be continued until the formic
acid level is low enough to prevent any significant hydrolysis during the
anticipated storage period.
The drying of the washed membrane fUm requires a technique that
takes into account the tendency of the membrane to shri~ linearly when dried
unsupported, wLth the Fesult that the dried membrane film is warped. In
~ ~601~07
order to obtain a flat uniorm film, the mernbrane must be restrained from
shrinkage during drying. One convenient way to do this, is to roll up a
continuous web on a plastic or metal core, with a high degree of tension so
as to obtain a tight roll, then ~îrmly wrap this with a rigid but porous outer
5 wrap, and then dry the assembly. Other methods of preventing shrinka~e,
such as tentering, or drying in drums under felt, are also satisfactory.
Individual membrane sheets of a selected size can be dried to
produce flat sheets free of warpage by clampinD the sheets in a frame
restraining the sheet from shri~age on all four sides, and then heating the
10 framed membrane at elevated temperature until it has been dried. We have
discovered that two or more equally sized membrane sheets can be placed
in contact a~d dried together in a frame $o prevent shrinkage. When this is
done, the contacting layers adhere to each other, and can thereafter behave
as though they were a single sheet. When the indi~idual starting sheets are
15 relatively thin, e.g. under 0.005 inch thick, and are of the ~supported
~subs~ate free) type, they ma~p be subsequently cut to size, for example,
by steel rule dies, and are thereafter for practical purpose~ a single sheet
or disc of filter medium.
The membranes can be dried in any of the ways described above,
2~ and then corrugated, seamed to provi~e a closed cylinder, and end capped.
We have discovered that this process can be greatly simplified, while
producing a superior product, by corrugating tbe filter medium while it is
still wet, together with upstream ~d dowllstream layers oE dry porous
material, this material being chosed to be relatively rigid, and not subject
~5 to no more than a small shrinkage during the drying operation. The
41
corrugated pack so formed is lightly compressed, so that the corrugations
are in firm close contact, while being held in a holding jig, preferably one
perforated to allow free access for heating and escape oE vapor, and placed
in an oven to dry. The resulting dried corrugated assembly shows only slight
5 shrir~age, and the corrugated polyamide mernbrane so obtained is free of
~varpage, with well formed smooth corrugation crests, and flat faces between.
When formed into a filtering element by side seaming and end capping, the
porous support layers provide flow spaces for access of upst;ream (dirty)
fluid and passage out o~ the element for downstream (clean) fluid.
If the filter cartridge is made using two or more thin layers of the
polyamide membralle, these will be firmly adhered to each othe} at the
conclus;on of the drying operation, and behave mechanicall~ as though they
were a single layer.
The control of the precipitation so as to obtain the formation of a
15 hydrophilic polyamide membrane sheet of desired flow characteristics and
pore size requires that the casting resin solution be controlledwith respect to a
characteristic referred to herein as "nucleation". The variables that must
be controlled include the choice of resin and of solvent and nonsolvent7 the
concentration of the resin in the starting polyamide resin solu~ion, tempera-
20 tures of all components, the quantity and mode of addition of nonsolvent,including rate of addition, intensity of mixing during addition, and the geome~ry
of the apparatus, the latter including especially size and location of the nozzle
through which the nonsolvent is added. For a given resin, solvent and non-
solvent, the effect of these variables on the degree of nucleation is qualitatively
25 stated in Table lll.
42
~6(~007
TABLE Ill
Variables af-fecting degree of nucleation
Dir ection oE change
to obtain a higher
Type of Variable Variable degree of nucleation
~.
Mixing conditions Temperature De~rease
Rate of nonsol~lent addition Increase
Size of inlet opening through
which the nonsolvemt is fed Increase
Distance of the inlet opening
from actual mixing area Increase
Intensity of mixing Decrease
Concentration of the ~c of resin Increase
components in the
15 casting solution ~c of nonsolvent Increase
_
Degree o~ nonsolvency
of the nonsolvent Tncrease
In Table Irr, the concentration of solvent is not included, as it is
defined by the concentration of the resin and the nonsol~ent.
It will be appreciated that the intensity of mixing in a g~ven system is
a function of a large number of variables. However, for a gi~en system the
relative intensity of mixing can be expressed in terms of the rotation rate of
the agitator~ or of the cutting blades of a homogenizer, etc. For a continclous
production system (as opposed to a batch operation) an in-line mixer is
requi~ed, and in a suitable designed multiblade ml~er about 1/4 to 2 hp is
required to produce about 30 kg per hour of 2000 centipoise casting resin
solution at a rotation rate between about 200 to 2000 rpm. Such equipment
43
0~ ,
can take diverse forms, and can take any oE a number of the designs commonly
used in the mixing art, since the various mixing principles can all lead to
similar results.
Because the intensity of mixing is diEficult to quantify, transfer of
5 manufacturing technology from batch systems, to continuous systems requires
trial-and-error experimentation, varying the operating condition parameters
~mtil one obtains the desired membrane sheet, all of which is within the
capability of one skilled in this art, since it involves manipulation of variables
that are cus$omarily adjusted in chemical process industry manufacturing
10 processes.
The importance of mixing intensity and oE the other conditions
related to mixing cannot be overemphasized. For example, a series of casting
resin solutions with the same concentrations of the same resin~ solvent,
and nonsolvent~ and the ~ame temperature and viscosity can be produced by
15 simply changing the mixer rpm. The most highly nucleated of these casting resin
solutions, made using the slowest mixer speed, will then produce a membrane
having an absolute pore ra$ing of 0.1 ,uM; the next more highly agitated
casting solution, cast into the same bath, will, if the mi~ing rate was correctly
chosen, produce a 0. 2 IlM absolute membrane, and similarly by using
20 successively higher mixing rates membranes can be made with absolute
ratings of 0.4 ,uM, 0.6 ,uM, 0.8 ,uM, etc.
The nozzle diameter through which nonsolvent is delivered during pre-
paration of the casting resin solution is also very important. It is at this nozzle
that the precipitate forms, which at least in part subsequently redissolves, and
25 the formation and complete or partial redissolution of the precipitate appears
44
307
to play an essential role in the preparation of the casting resin solutions of
this invention. With all other parameters maintained equal, a casting resin
solution of quite different characteristics, in terms of the pore size of the
resulting membrane, will be obtained by simply varying the diameter of the
nozzle. We have used nozzle diameters varying from 0. 013 mch to 0.125
inch diameter, but smaller or larger nozzles could be used with successful
results.
Not only can a casting resin solution of given composition and tempera-
ture be made by varying the mi~ing intensity and thereby the degree of nucleation
10 to produce greatly different membranes, but the converse is trueS namely,
membranes of equal or nearly equal characteristics can be made usin~ a t5; ide
variety of resin7 solvent, and nonsolvent concentration in the casting resin
solutionj for example, an increase in water content will increase the degr ee
of nucleation, but if the mi2~ing intensity is also increased~ a casting resin
15 solution will be obtained with the degree of nucleation unchanged, and the
membrane cast from this casting resin solution will ha~e characteristics
equal to that made from the lower water content casting resin solution.
The relationship between the degree of nucleation and the a~solute
pa~ticle removal rating of the resulting membrane is graphed in Figure 1,
20 which shows an inverse relationship between the pore diarneter of the
membrane sheet and the degree of nucleation, i. e. 9 to obtain small pore
diameter, a high degree of nucleation is required.
Reference to the gx aph of Figure 1 shows that in Region A, where
the degree of nucleation is very small, the pore size tends to become
25 nonreproducible. In addition, the pressure drop at a given pore diameter is
07
high. Membranes made on the assumption that the concentrations of the
components are the controlling factors, and without nucleation, for example,
by the process of Marinaccio, fall into this range, and tend to be
of relatively poor quality. In Region B, the pore size decreases in a regular,
5 though not necessarily linear fa~hion, as the degree of nucleation increases.
In Region C, the casting resin solution becomes increasingly populated by
particles of resin which have not redissolved, but still produces good
quality membrane if these are remoYed by filtration prior to casting; and
in Region D, the resin solution i~om which these lumps ha~e been removed
10 by filtration becomes unstable, and prone to early local or orerall gelation
before the film can be cast. The very high degree of nucleation in Region D
is sometime9 manifested by an opalescent ~ppe~rance, su~gesting that the
nucleation procedure has resulted in an e~cessive number and/or e~cessively
large nuclei.
Because methods of achieving a required intensity o ml~in~ vary
so greatly among the various types of equipment used in the mixing aLrt, it
is not possible to quantify this characteristic. Consequently, any given
apparatus must initially be used on a "cut and try" basis to produce ~asting
solutions o~ the desired characteristics, applying the principles tabulated
20 in TablelII. Once the parameters of mixer rate, concentration~, tempera- -
tures, flow rates, etc. have been es~lblished, casting resin solutions having
quite reproducible characteristics can be produced in the B and C 3~egion o-f
Fi~ure 1, on successive days or weeks of operation.
A favorable condition for producing membranes having low pressure
25 drops and particle remoYal ratings covering a wide range utilizes a starting
resin containing 15.5~c of 42,000 molecular weightNylon 66 resing 83.23~C
46
~J~ 3!07
OI formic acid, and 1. 27tC of water . When this starting resin solution is diluted
using the conditions of Examples 1 to 39, the results obtained in Figw7ce lU
are obtained. The product range of I~L is such that membranes are obtained
with absolute particle ratings ranging from about 0.1 micron (for e~ample, a
O. 012 inch thick membrane with KL ~ 100 psi) to about 1 micron (for example,
a 0. 001 inch thick membrarle with KL ~ 27 psi).
The curves of Fi~ure 10 were obtained usin~ a specific in~line
mixer configuration, in which the rotor was 2-1/2 inch in diameter. The
same results can be obtained by using other mixers, and theRPM needed
to produce these results rnay vary; however, it is within the ability of a
person familiar with the art to determine by test the conditions required with
his apparatus to duplicate the intensity of mixing reprssented, for example, hy
the 1950 RPM and 400 RPM conditions of Figure 10, and once this has been
accomplished, the conditions for making mem~ranes covering the whole
range of Figure 10 will be apparent to him.
This same correlation of mixing cvnditions would then be equally
applicable to the other Examples of this invention, m which an in~line
mixer was used.
The casting resin solution can be extrudedabove or under the surface of
the nonsolvent bath, especially if used to make hollow fibers; this process
is more easily realized in practice by using relatively high resin viscosities
(e. g. 100, 000 cp) and rapidly setting casting resin solution in relatively
lower formic acid concentration baths, e. g. in the 20 to 40~tc range.
As previously described, three types of substrates are used:
~7
. ,
~ o~ .
(a) nonporous, for example, commercial polypropylene or other
plastic film~ glass, etc.;
(b) porous, not wetted by the casting resin solution; and
(c) porous, wetted by the casting resin solution.
The nonsolvent precipitation baths used in this invention contain
a mi~ture of solvent and nonsolvent for the resin. The characteristic of the
bath which has an important effect on the properties of the resulting membrane
is the relative concentration of solvent and nonsolvent in the bath. If the con-
centration of solvent is ~ero, or at low level, Eor example, below 20~C, a heavily
10 skinned membrane will be obtained~ If the concentration is adjusted to one of
the preferred ranges of this irlvention (about 43 to 55~ of formic acid, in the
case of a bath containing only water and formic acid) the reæul~ing membrane
has uniform pores from one face to the other.
If the bath concentration is 43 t~ 55~c, and the su~strates used are
1~ of types (b) or (c) described above, the pores will always be uniform through
the thickness of the polyamide membrane. However, if the film is cast on
a nonporous substrate of type (a), it is important that the substrate surface
be wettable by the casting resin, and by the bath fluid. Glass, and similar
surfaces, are naturally so wetted; however, synthetic plastic film materials,
20 such as polyethylene, polypropylene, polyvinyl chloride, andpol~ester are
not, and if the casting solution is spread on such a substrate, and immersed
into a 45~'c formic acid 55~c water bath, it will form a film with open pores
on the face in contact with the bath, the pores being uniform throughout most
of the body of the film1 but with a dense skin on the substrate side. We have,
25 however, discovered that if such plastic films are rendered more wett~ble,
48
O~
- for example, by surace oxidative processes such as chromic acid ~eatment
or corona discharge treatment, the resulting membrane is skinless on both
faces, and of uniform p~re size throughout. In such a membrane, it is
difficult if not impossible to determine by any manner of appraisal which
5 side was in contact with the substrate.
To obtain such skinless membrane sheets, a wide range of surfaces
can serve as the substrate, provided that the critical surface tension is
maintained at a sufficiently high value. This wi~ll vary somewhat depending
on the concentration of formic acid in the resin solution and in the bath, and
10 the temperature, -and is best determined by trial-and-error treatment of -
the substrate surface for a given system. Critical surface tensions required
are generally in the range from about 45 to about 60 dynes/cm, and most
often in the range of from 50 to 56 dynes/cm.
If a given casting resin solution is immersed as afilm into a series of
15 baths, each with slightly increasing water content, the characteristics of the
men~brane on the side facing the bath will gradually change, producing film
which have finer pores at and near this face, compared with the balance of
the thickness of the membrane. These finer pores show a gradual transition
- into the uniform pores of the balance of the membrane. Such membranes
21~ are described herein as "tapered pore membranes", and are useful in that,
when filtering some suspensions, with flow from the coarser to the finer side,
longer service life (higher dirt capacity) is obtained, with equal removal.
_gure 7 shows scanning electron micrographs of a tapered pore membrane.
The bath sol~ent concentration required to obtain any desired taper pore
25 membrane varies consiclerably, depending, for example, on the state of
49
o~
nucleation oE the cas-ting resin solution, and shoulcl be determinedfor agiven set
of condition~ by tria~and-error; however, in the case of a water-formic
acid bath, it is never less than 15 to 25/3tC, and usually is near to 30 to 35~c
of formic acid.
As the bath water concen~a.tion increases, the membranes begin to
form with increasingly heavier skins, and are characterized b~r high pressure
drop, and poor pore size distribution characteristics.
The uniform pore membranes made by the process of this inYention'
such as those shown in the scanning electron micrographs of Figures 5 and 6,
10 are characterized by liquid displacement curves such as shown in Figure 3.
When the membrane is immersed into water, its pores. are filled by the
water- forming within the membrane a film of immobilized water, wh~ch
remains in place when the membrane is removed from immersion. When
air pressure is then applied across the membrane, there is noted a very
5 small flow of air. This air flow when divided by the applied air pressure
re;nains constant as the pressure is increasedS when plotted as in Figure 3.
From the thickness of the film, and the known di~fusion constant of air in
water, it can be calculated using Fic~'s law, that this 1OW is due to diffusion
of air through the water film, and does not indicate flow through pores of the
20 fater medium. At a sufficiently high pressure7 the flow as plotted in Figure 3
is seen to increase suddenly, reflecting displacement of water from the
~argest pores, and flow of air through these pores, and the curve becornes
nearly vertical. The sharpness of this rise will be appreciated by noting
that in this region, the membranes of this invention require less than a 1~c
25 to 3~Zc increase in pressure drop to accomplish a 5000 fold increase in air
flow rate.
The rapid transition from zero -flow of air (except that due to
difusion) to a v~ry steeply rising rate of Elow for small changes In applied
pressure, characterizes uniform pore media, which have sharply defined
removal characteristics; such mediawill, for example, quantitatively
5 remove one bacterium, but will allow an only slightly smaller organism
to pass. ~uch membranes generally also have favorably low pressure drop,
for a given removal.
Skinned membranes behave very dif~Eerently; when water wetted
and their air flow-pressure drop relationship is determined, the curve is
10 not flat initially, but slopes upward, indicating presence of large pores;
transition to a more nearly vertiral line is slow, with a large radius, and
in the "vertical" area, instead of the sharp rise of ~p~ure 3, a sloping line
is obtained, reflecting a wide pore size range. Such membranes are poorly
suited to obtain sterile fil~ates when challenged by bacteria; either a
15 nonsterile fluid is obtained, or if sterility is gotten, it is at the cost of
very high pressure drop to achieve a low throughput rate~
It is apparent from the preceding discussion that control within
narrow limits of the concen~ation of formic acid in the nonsolvent liquid
in the bath is desirable to obtain a uniform product. In a continuous process,
20 this control is obtained by an appropriate feed to the bath of nonsolYent liquid,
while simultaneously withdrawing some of the bath liquid to maintain constant
total bath volume. A relatively higher concentration of formic acid enters
the bath from the casting resin solution, and the concentration of formic
acid in the bath therefore tends to increase. Water is therefore constantly
25 added to the bath to compensate. Accordingly, control of the rate of
51
~L~363 D7
addition of water and of the rate of withdrawal of surplus bath solution will
give the desired result: substantially constant concentration of formic acid
in the solution, within the limits that give a membrane of the characteristics
desired.
Thus; Example 47shows that in order to obtain a skinless membrane
sheet havin,, a uniform pore distribution, with fine enough pores to quanti-
tatively remove all incident bacteria and particles over 0. 2 ,uM, a relatively
highly nucleated castin~ resin solution is cast as a film and the membrane
precipitated in a 46. 4~c aqueous formic- acid solu~ion as the bath liquid.
To produce a membrane with tapered fine pores, a film of less
highly nucleated casting resin solution is precipitated in membrane ~orm by an
aqueous 25~C formic acid solution as ~he bath, as in Example 60.
It is instructive to note that in the range o-E 0. 2 ,uM and below, the
uniformity from face to face of commercially available regenerated cellulose
and cellulose ester membranes becomes quite poor, and such membranes are
to some degree tapered pore types. In the sarne range, the membranes of the
invention remain uniform, or ma~r be tapered, as desired.
Thus, in the continuous production o~ membrane sheets in ar-cordance
with the invention, to obtain uniform ch~acteristics in the membraneS the castinD
20 resin solution must be prepared under carefully controlled conditions and
the bath liquid compo9ition must remain constant. Such a liquid is referred
to as an "equilibrium bath'~, i. e., a bath in which the concentratlon o~
ingredients remains con5tant~ regardless of additions and withdrawals.
To illustrate, consider a casting resin solutioncontainingl3~c resin and
25 69~c formic acid with the balance water, continuously being cast in film
52
~oool~
form on a substr~te, and then plunged into an aqueous nonsolvent bath
containing 46~C formic acid. As the resin membrane precipitates, a pro-
portion of the solvent from the film of casting resin solutlon ~Nhich contains 69
parts of formic acid to 18 parts of water, or 7~-3~c formic acid) diffuses
5 into the bath, thereby altering its composition. To counteract this, water
is continuously added to the bath at a rate conltrolled, for example, by a
device using density measurements to report ~Eormic acid concentration, at
the 46~7C level, and bath liquid is withdrawn continuously to tnaintain total
bath volume constant. MaiIItaining this equilibrium bath makes it possible
10 to continuously produce a membrane sheet having uniform pore character-
is~ics.
When used continuously, at high production rates, the bath temperaturewill gradually increase; cooling by a heat exchanger ma~T be used to maintain
constant conditions.
From the above-mentioned casting resin solution and bath, unsupported
membrane sheets can be made by casting the resin solution onto an endless
belt, or onto a plastic sheet unreeled from a roll, as a substrate to support
the cast film.
The membrane sheet has a tendency to adhere to the subs~ate
surface on drying, and it i~ therefore important to remove the membrane
sheet from the surface while it is still wet, and before it has been dried
and developed adherency.
Unsupported membrane sheets obtained by the process of the invention
are quite strong, with water- wet tensile stren~ths in the rage oE 400 to
25 600 lbs/sq. inch, and elongations generally exceeding 40~c.
53
~.G~oO7
For some applications, even higher tensile strengths may be desired.
In additi~n, unsupported membrane sheet requires special care to manipulate
in the typical range of thicknesses from 0. 002 to 0. 010 inch in which it is
normally manufactured. In such cases, a supported membrane sheet is
5 desired. Such membrane sheet is prepared by forming the film of resin
solution on a substrate which adheres to the membrane sheet after it has been
precipitated thçreon. Either of the two types of substrates can be used; those-
which are not wetted by the resin solution, and those which are.
The unsupported filter ~nembrane obtained at the conclusion of the
10 !nembrane forming process is wet with water, also contains a small amoun~
of residual formic acid. This product can be dried in various ways.
It can, for example, be collected on a roll on a suitable core in
len~ths from 50 to 100 linear feet and placed in an oven until dry. Durir~
drying, some shrinkage occurs, but an acceptable product is obtained.
It is also possible to clamp a léngth OI membrane in a frame holding
all sides against shrinkage, and then dry the membrane by e~posure to heat,
as by infrared radiation, or in an oven in air. The resulting sheet is very
flat, and when discs are cut from it, these are adapted for use in apparatus
designed to accept disc filter membranes. The membrane discs are quite
20 strong and flexible, and can be readily and reliably assembled in such appa~atus.
A similar product can be obtained with less hand labor by passing the
wetted membrane sheet Dver a hot drum, against which it is firmly held by a
tensioned felt web or other porous sheet, and the dry web collected as a roll.
If two or more layers of wet unsupported membrane sheet are dried
25 in contact with each other~ using any of the drying methods described above,
54
;~ a6~0~
they adhere to each other, forming a multi-layer structure. No honding
agent or other adhesion technique is required.
The resulting multi-layer membranes are useful in the manner of a
single layer filter membrane. SinGe in manufacture a small proportion of
5 undetected faults may occur, caused,for example, by bubbles of air entrained
in the casting resin solution, using two layers instead of one neutralizes
such areas, covering them over with a second layer of filter membrane that
is also capable of providing the required removàl rating; an extremely high
degree of reliability is obtained in this manner.
Very good adhesion o adjacent layers is also obtained if a layer of
supported resin membrane and one not supported are dried in contact, using
the same procedures. In this manner, filter media can be made in which
a supported layer of uniorm poxe size is bonded to an unsupported t~pered
pore membrane layer, which provides e-fficient prefiltration. The fine face
15 of the tapered pore layer would be about the same pore size or somewhat
larger than the pore size of the supported layer, and this face would be
adjacent to the unsupported layer.
Supported filter membranes in accordance with the invention are
particularly well suited to use on filter presses, where self-sealing character-
20 istics are needed, and the filters are subjected to large stresses. They arealso useful in making plain, or corrugated filter cartridges for use at high
differential pressures, or for impulse type æervice .
The filter membranes of the invention are well suited for use as
the filter media in filter cartridges. Filter cartridges are self-contained
25 filter elements, provided with a filter sheet in tubular form capped off by
.
~6~ 07
end caps at each end. Either or both end caps can have a throuDh opening
~or fluid circulation through the filter sheet in either direction. Filter
cartridges are designed to be installed in and to be readily removable from
filter assembly housings when replacement is necessary.
A good filter cartridge has a filter sheet that is free of faults, and
with removal characteristics that are relatively uniform with stated standards.
Filter cartridges take many forms, including simple cylinders, corrugated
cylinders, stacked discs, etc.
Of lthese configurations, a favored form for the filter membranes of
the invention is a corrugated cylinder. Such a cylinder is made by corrugating
one or more layers o~ supported or unsupported wet mernbrane (two layers is
preferred) sandwiched between two open porous or foraminous sheets which
provide for fluid ~low up and downstream of the contacting surfaces of the
filter medium within the corru,,ations. The resulting corrugated structure
can be dried while lightly restrained, in the course of which contacting
membrane layers are bonded together, thus forming a more rigid, stronger
structure, and then seamed closed along the contacting ends, using heat-sealing
techniques similar to those used for sealing conventional thermoplastic filter
materials. End caps are then attached in a leak-tiDFht manner to each end o~
the resulting cylinder. The preferred method is in a~cordance with U. S. patent
No. 3, 45'?, 339, patented December 8, 1965, to Pall et al. The end cap
material can be an~ of a wide range of thermoplastic synthetic resin materials,
particularly polypropylene, polyamides, polyesters and polyethylene. Polyester
end caps, particularly polyethylene terephthalate and polybutylene terephthalate,
seal very well to polyamide membrane materials, and have the advantage that
5Ç
~1~6~
the assembled carh idge is rapidly wetted by water, permitting a test using
the standardized procedures of the invention to verify the integrity of the
assembled filter cartridge.
In the manufacture of corrugated cylindrical filter cartridDes, a
seam must be made joining the ends o-f the corrugated structures. Since
the polyamides used to mal~e the membranes of this invention are thermo-
plastic, heat sealing may be used to close the seam, and is for many or
most purposes an acceptable method. Heat sealing does have some
disadvantages, however:
(a) in order to make the seal, it is, practically, necessary to
bend the last leaf of each outerrnost corruga$ion to an angle of
90, which is sometimes difficult to accomplish without
weakening or other injury to the filter medium at bend;
(b~ the temperature used and duration of the sealing operation
need to be changed to accommodate changes in thickness of
the filter medium layers usedj and
(c) a weakening of the structure occurs due to the introduction
of a stress concentration at the edge of the seal a:rea; ~ highly
stressed, the filter will fail at this edge, in preference to any
other part of the assembly.
All these disadvantages are overcome by a novel joining technique.
We have discovered that a solution of trifluoroethanol containing 3 to 7~c
of Nylon 66 in solution can be applied to the outermost face of each end
corrugation, and the two surfaces then lightly clamped-together, and the
fluoroethanol allowed to evaporate. Other solutions may be used, for example,
67
6~
a 33~c solution of Nylon 66 in formic acid, similarly solutions of polyamide
resins in hexafluoroisopropanol or he~a-~luoroacetone sesquihydrate. An
excellent seal results, free of all the disadvantages enumerated above; indeed
the seal area is now s~onger than ~he remalning corrugations.
The quantity and concentration of the resin solution are quite
noncritical, and good seals have been made with as little as zero percent
or as much as 9~ of Nylon B6 resin in the ~ ifluoroethanol solution9 but Ul
this solvent solutions in the neighborhood of 5~Yc are preferred, being sta~le,
and having a convenient viscosity if a high molecular weight resin is used to
10 prepare the solution. Solution9 in formic acid have also been successfully
used.
The accurate determination of effective pore size for membrane filter
media that is meaningful in its representation of expected e~ectiveness as
a ilter is difficult. When a uniform pore filter medium of this invention,
15 or any of the currently marketed uniform pore membranes are examined
using a scanning electron microscope~ e. g. as is shown in Figure 5? and
the apparen~ pore openin~s as seen on the micrograph are measured, a
pore size is determined which is about three to five times the diameter OI
the largest particle which the filter wlll pass, as determined~for example,
20 by bacteria challenge. Similarly, it has been attempted to a~certain the
pore diame~er from the KL valu~ as determined by the procedure of
applying air pressure to a wetted element, obtaini~g the KL value in the
manner ~hown in Figur0 3, and inserting the so determined pressure into
the well known capillary rise equation; when this is done a diameter is
25 determined which is a~out four times the absolute removal rating of the
filter medium as determined by bacterial challenge.
~8
~16~0~7
Such methods, upon reflection, appear to haYe little relevance to the
capability of the membrane as a filter. What the user needs to know is not
pore size; rather, it is the capability of the filter in removing particulates,
bacteria, yeast, or other conta~ninants.
Contrary to established thinking, we have determined by test that the
effectiveness of membranes of structure silrlilar to those of this illvention asfilter media is dependent not only on pore size, but also on thickness. In the
development of the present invention, it has, for example, been demonstrated
that of two memb.; anes, one having small pores and quite thin, and the other
having relatively larger pores and much thicker, the membrane have the
larger pores but the greater thickness may be more effective as a filter.
Accordingly, the effectiveness of the membrane sheets in accordance
with the invention as filter media is rated not in terms of pore size, but in
terms of effectiveness in the remo~ral of a contaminant of known dimPnsions.
One of the principal applications of this type of filter membrane is to deliv~
a filtrate freed of all incident bacteria, hence bacterially sterile. A technique
usually used in the industry to determine the ability of a filter to deliver
bacterially sterile effluent is to challenge it with a suspension of Pseudonnon_s
diminutiae, which is a small diameter relatively nonpatho~,enic bacterium
referred to in abbreviated form by Ps. Filters which successfully meet such
a challengre are generally accepted in the industry as being 0.22 micrometer
absolute in filter rating, and in any event Pseudomonas diminutiae is a
bacterium that represents the lower limit of bacterial dimensions. If no
combination of challenge conditions can be found which will allow even a single
organism of Pseudomonas diminutiae to pass, the filter can be regarded as
capable of quantitatively removing all bacteria.
59
()7
This invention employs a standardized test based on Pseudomonas
diminutiae removal that correlates such removal ~vith air flow measurements
through the wetted membrane and the thickness of the membrane, and is
capable of providing a quite complete characterization of the rernoval
characteristics of the membrane filter sheet being tested.
The removal of Pseudomonas diminutiae is a function not only of
pore size but also of thickness, and is expressed by the exponential
relationship:
TR TE~l 9
or log T~ ~ t log TR
where
TR is the titer reduction for the membrane and is the r~itio of
Pseudomonas diminutiae content in the influent to the content thereo in the
effluent;
TR the titer reduction achieved by a membrane of unit thickness; and
t is the thickness of the membrane.
As an example o~ the application of this formula, if a given membrane
has a titer reduction of 105, two layer~ of the membrane will have a titer
reduction of 101 ~ three layers of 1015, etc.
Since the incident test bæterium is monodisperse (i. e. ~ of uniform
dimensions), the applicability of this formula is self-evident. Its carrectness
has also been confirmed experimentally, by determining titer reductions for
1, 2, 3, 4 and 5 layers of the same membranes. As shown in ~
the resulting plot of log TR vs. number of layers is linear, as predicted
by the formula.
~L~60V()7
It is known in the industry to measure air flow rates through a
membrane which has been wetted by a liquicl; such measurements yield
useful information on the pore size characteristics of the membrane. We
have used, in the course of this invention7 a parameter designated as KL.
KL is a form of abbreviation for the "knee location" of the curve of Figure 3.
When the air flow/unit of applied pressure through a wetted membrane is
plotted against increasing applied pressure, as in Fi~ure 3, the initial air
flow is very small and the flow per unit of applied pressure remains nearly
constant, until a point is reached where a very small increment in pressure
causes a very sharp rise in flow; such that the curve becomes nearly verticalr
The pressure at which this occurs is designated as the KL for the membrane.
KL has been measured for a group of forty~five membranes made by
the process of this invention from polyhexamethylene adipamide; these
membranes were selected to cover a range of thickness from 0. 0015 inch to
0- 012 inch, and with a wide range of pore diameters. These same membranes
were then challenged with a suspension of Ps bacteria, and the num~er of
in~uentbacteria was then divided by the number of effluent bacteria, ~hus
determining the TR for each of the membranes. Thickness of eacb membrane
was then measured, in mils (one mil - 0. 001 inch),and using the formula
log T~ = t log TR
log TR was then calculated for each membrane, T bein~ the theoretical titer
~1
reduction for a 1 mil membrane.
KL values were measured, for both relatively coarse and relatively
fine membranes, for a number of thin membranes. These same membranes
were then laid up as 2, 3 and more layers, and the Kl~, values again measured
for the multilayers. In this way, a relationship between the thickness and KL
61
value of equal pore size membranes was determined; this relationship is
summarized in Table 1. IJsing Table 1, the KL values of the 45 membranes
were corrected to the KL which would apply to an equal pore size membrane
0. 005 inch (5 mils) thick; these values are dlesignated as KL .
Log TR for each membrane was then plotted against KL for that
membrane. All the results fell close to a single line, which is shown in
Figure 4.
Using Figure 4, the titer reduction (TR) which can be expected to be
obtained with any membrane made of hexamethylamine adipamide by the
lO process of this invention, can be calculated, using the measured Kh and
thickness ~t) for that specimen. The procedure is as ~ollows:
(1) measure the KL and thickness for the specimen;
(2) use Table 1, determine KL
~3) use ~L to determine TR i~rom Figure 4; and
(4) calculate TR from the equation TR = TRt .
There is an upper limit to the number of bacteria which can be
collected on a membrarle; by the time that about 1013 Ps. have been collected
per square foo~ of filter medium, flow through the filter has fallen to less
than 0. Ol'~C of a normal starting flow rate of 2 to 5 liters/minute per square
20 foot. This has been determined, by actual test, to be true for the membranes
of this invention, as well as for commercially available membranes, for the
full range of TR from 10 to ~103~. Thus, the figure of 1013/square Eoot may
be taken as a practical upper limit of invading Pseudomonas diminutiae.
This upper limit is taken in combination ~ith the calculated TR to
25 obtain assurance that a given membrane will yield 5terility under all conditions
~2
of use. For example, a membrane may be selected with an estirnated TR f
10'3; statistically, if challenged with 1013 PseudomonaS ~iminutiae, such a
membrane would have to be so challenged for 101 (or lO billion) times7 in
order to produce a single effluent wi~h one bacterium, and such a high ratio
5 may be taken as adequate assurance of sterility, hence the filter canbe
considered to have an absolute removal capability at 0. 2 ,uM. In practice,
it is difficult to consistently produce a membrane with an estimated TR f
exactly 10~3, but it is feasible to establish a permissible range, say 1023 to
1027, with 1023 as the lower limit, and thus obtain assurance of consistency
lO bacterially sterile filtrat~s.
In a similar fashion, KL and thickness can be correlated with titer
reduction for larger bacteria7 yeasts of known size, and other particulate
material, the Iatter assayed by particle detection methods, over a size
ran;,e from under 0. l IlM or larger.
The curve of Figure 4 is applicable to the membranes made by the
process of this invention. The process by which this curve was developed~can
be applied to membranes made by other processes. The location of the curve
for other membranes may shift somewhat, but we have done sufficient testing
using curre~tly marketed-uniform pore dry process membranes to determine
20 that the same principles are applicable.
The horizontal portion of the curve of Figure 3 is truly horizontal only
if the pore size is quite uniform. Uniform pore media are further character~
ized by a sharp change in slope to a nearly vertical course at the ~L value.
If the filter medium is relatively nonuniform in pore size, it will tend to have
25 a distinct slope in the ho:rizontal portion of the curve, and exhibits a relatively
.
63
~ 6~007
large radius for the change in slope to the more vertical portion of the curve,
followed by a sloping rather than a nearly vertical portion.
The lower or horizontal portion of the curve is a measure of the
diffusion of air through the immobilized, liquid film which fills the pores of the
5 membrane. The wetting liquid may be water, in which case a relatively low
air flow is obtained in the horizontal part of the curve, or alcohol, in which
case the diffusional air flow is higher. At the change in slope7 the wetting
liquid begins to be expelled from the pores, and in the vertical portion of the
curve, a large number of nearly equal size pores begin to pass air .
When the data of Figure 3 is plotted for a tapered pore membrane,
that is, one with larger pores-at one face, taperingr to a smaller pore at the
other face of the membrane, the curves obtained by reversing $he direction
of pressurization do not coincide. Instead, two distinct curves are obtained?
one Ilat, and the other higher and sloping upward, of which the sloping curve
15 with higher flow values is obtained when the more open side is upstream, and
reflects the penetration of air partly into the coarser face of the membrane,
thereby effectively decreasing the thickness of the liquid film, and hence
increasing the air diffusion rate.
Thus, by applying air pressure and measuring; flow throug~ a
20 membrane successively in both directions, it is possible to determine wbether
it is a uniform or tapered pore membrane. If the flow-pressure curves are
equal, or nearly so, in both directions, the pores are uniform, and the
method describ`ed herein for relating KL and thickness to titer reduction for
any given organism, or to a rnonodisperse particulate, may be applied to
25 that membrane.
64
.... .
07
The following Examples in the opinion of the inventor represent
preferred embodiments of the invention:
E~AMPLES 1 to 5
.
Nylon 66 resin pellets of molecular weight approximately 42, 000
5 were dissolved in 98. 5% formic acid, to yield a 35C solution containing
15. 5% of the resin. Without delay, this solution was delivered at a now rate
of 250 g/minute to an in-line mixer. Simultaneously, a controlled water flow
at 31C was delivered to the mixer, the quantit~ being such as to produce as the
e~fluent a casting resin solution containing70. 2% of formic acid and 13.1'3~ of
10 the resin. The casting resin solution was filtered through a 10 ,uMfUter to
remove visible resin particles, and wa~ then formed as a thin film by a doctorlng
roll with 0. 0085 inch spacing on a moving polyester sheet surface, which had
been pretreated by corona discharge to improve its wettabilit~, and in less
than 3 seconds immersed into a bath containing 46. 5% formic acid) balance
15 water, for approximately 3 minutes. Ba~h concentration was maintained
constant by adding water continuously, in the amount required. The nylon
membrane so ~ormed was washed with flowing water for 1 hour. Two layers
of the membrane were removed from t~}e polyester substrate sheet and ove
dried in contact with each other, while restrained to prevent shrinkdge of
20 the length and wiclth during drying.
The rotation rate of the in-line mixer was varied from 400 to 1600 RPM
during this run Table IV shows the product characteristics obtained. In
this Table, "uniform pores" means that the pore size was equal, as determined
by SEM examination throughout the whole width of the membrane. Examples
25 1 and 2 represent conditions in region A of ~igure 1, in which the degree of
31 ~6~
nucleation is too low to produce a satisfactory product; in this zone pressure
drops are high, and product characteristics tend to be nonreproducibleD
Example 5, in which mixer speed was 400 ~PM falls into region D oE
Figure 1, and resulted in an unstable condition, with so much precipitating
5 resin generated within the mixer, that it began to clog, such that casting
resin solution could not be delivered.
The wide variation in behavior and in product characteristics, for
the same casting resin solution as defined by the concentration of its
somponents, should be noted.
XAMPLES 6, r? and 8
Casting resin solution was prepared and processed as for E~ample 4,
except that it was heated by means of an in-line heat exchanger to respectively
53, 61 and 68C prior to casting. The product characteristics were not sig-
nificantly different from those of Example 4. This result confirmed previous
15 test ~lata indicating that temperature ~f the casting resin solution is not a
signific~t parameter, except insofar as viscosi~ may be reduced to the
point (below about 500 cp) where casting problems may be experienced.
EXAMPLES g to 13
The membranes were prepared in the same way as Examples 1 to 5,
20 except that the quantity of water added was such as to produce a casting resin
containing 69. 8~c of formic acid and 13. ~c of resin. The results are shown
in Ta~le V. The casting resin solution made at 1950 RPM mixer speed was
insufficiently nucleated, resulting in a pOOl product with high pressure drop.
66
~.6a~07
~MPLES 14 to 18
,
These membranes were prepared in the same way as Examples 1 to 5,
except that the quantity oE water added was such as to produce a casting resin
solution containing 69.0~C of formic acid and 12.85~C of resin. The results
5 are shown in Table VI.
E~?l,ES 19 to 39
. . _ .
These membranes were prepared in the same way as Examples 1 to 5, .
except that the quantities of water added were such as to produce casting resin
solutibns containing 71.4~c, 67-5'3~c and 66.0~C of formic acid, and 13.3~ZC, 12.55~C
10 and 12. 41% respecti~Tely of resin.
The results are shown in graphical form, along wi~h the data of
Examples 1 to 19, in Figure 10. Figure 10 includes only those membranes
which fall in the regions B and C of Figure 1, and there~ore have favorably
low pressure drop in proport.ion to their thickness and particle removal
l5 capaoility, and -e con-islently reproducible.
. ' - , '
67
.,- Q
d d d
O o ._
td ~} h ~ f~ 07
~ s:
r3 j~ ; h h ~3
K
~ o
. ~ ~ o o
h
O
.
E
¢ ¦ ~ _I If~ ~ O
~ O tD
<I' ~
' t~
P,
. ~ .
O
. ~q
.q
h ~ O O ~;
O g O
. . ~;
a) :~
X ~1 IN c~ h
lY I
68
. . .
~6ai~7
~:
o
a)
~ U~
~' O O O
h ~ O O ~ ~
P; ~i P
:
~ a~ N r-- ~ ~ ' ' '
o~ o o C~ o
. . a~
P ~ ~~ ~ O
I ¢ ~
¢ ~1 z; ~ CD A A
~ . t~ . '
-~ ~ ~ CD ~ O~ h
. ,~ M C- ~b
o
. ~ .:
~D ~ ~ ' O
W P~:
. ~ . .
' ' ' ~LO V . ~
.~ ~ o ~D00 ~ O
r- ~DL~ ~ ~
~C~ ooooo .~
.~ ~, ~ o o o ~ ~
K cr~
~:4 . U3
~ O Cl~ O ~ C~
C~ ~; ~ ~ ~ ~ P~
,~1
,~, . .
6g
o ~ ~60at~17
.
C~ ~
C) ~ O
~; o o o
~n ~ ~ ~ ~ s~
0 C~ U~ o o C"
~- ~ o o o o o
. ~ :
. , ~ .
~5 o o C~ - ~
~ ~ P~ I ~ o o 4~
.~ , 1 ~ ~ ~ ~ ~ ~
_, ~ A A o
, ~ o CO ~ ~: :
~3 ~
~ o~
~ ~ c~
ua ~ ~ ~ a~ ~ `
. ~ ~~t ~
. ' ~ .
Ul ~ C~ ~
~ r~
' 5-~ .
, ~ CD O ''
g P:i r~
. ~ .
S~ ~ O O O ~ O ~ '
X U~ O O O O ~~
-~ F4 ~ Cl~ CD Q.
~;
O' d~
~; ~ ~ ~1
~1 . .
~ .
. . .
0~?~)7
E~AMpLES 40 to 46
These membranes were prepared in the same way as Examples
1 to 5, except
(A) Starting resin containing 14. 5~O of Nylon ~6 was delivered to
5 the mixer at 400 g/minute.
~) Water was added in various quantities, to obtain formic acid
and resin concentrations as listed.
- ~C) Doctor roll set at 0O 022 inch.
The results are listed in Table VII.
ql
007
D O ~ ~ C~ O O O '
o ~, o o ~o
~ ~ ¦ O O O
~ /\ ~ ~ ~ . ~
~ CD ~ ~ ~ O ~ ~
. ~ ~ o~ . o
~1 ~1 ~~ . ~ O O a~
~ ~ .~ ~ ~ ~ ~ ~ ~. C~ .
~ $
o~ o C~
,~: . h
~.q ~i _I N C`;l C;i C;l C~
K
, C~
O ~;
~ r~
~ .
o . o
o In ~ ~ ~ ~ ~ ~
5~ . .~ :
~ ~ o o o o o o o ~ .
. p,, o o o o o o s.,
r; 0 cn o~
a~
o o
C~ :Z ~ ~ ~ ~ ~ ~ ~ ~
W : :
~ o~
q~
ao7
E~AMPLES 47 to 5 0
.. :
Nylon 66 resin pellets of molecular weight approximatel~r 42, ooo
were dissolved in 98. 5~3~ formic acid, to yielld a 35C solution containing
15. 5~ of the resin. Without delay, this solution was delivered at a flow rate
of 250 g/minute to an in-line mixer rotating at 1200 RPM~ Simultaneously,
a controlled water flow also at 30C was delivered to the mixer, the quantity
being such as to produce as the effluent a casting resin solution containing
69. 0~ of formic acid and 12. 9~ of resin. Temperature of the resu]ting casting
resin solution was 57C. The casting resin solution was without delay filtered
10 through a ~0 ,u M filter to remove visible resin particles~ alld was then formed
into thin films by a doctoring blade with 0. 010 inch spacing on glass plates,
and in less than 10 seconds immersed into a bath containin~ formic acid and
water, for appro}~imately 5 to 10 minutes. The nylon membranes so formed
were washed with flowing water for 1 hour. Two layers of the membralle were
15 oven dried in contact with each other, while restrained to prevent shririkage
of the length and width during drying.
Table vm shows the product characteristics obtained, for various
bath concentrations.
73
a~
~D h o
h U~
1 ;~ P
: - ~
~ o ~ C~ O
~ ~ ~ O O O O
~ . O . . O
.~C s ~ I K K " o S
o~ ~ ~ A A ~,
. .s~
~ ~ ~ o o
u~ ~ C~ ~ ~ CD , h
: ~ ~ ~ e~ ~ ~ ~3
O'
~n
~:s o ~
._. ~ . ~
~ o ,~; ; ~D ~ C,~ N ~
a) :1
Q. ~ h
P~
~1 -'5
~ .
74
X~MPLES 51 to 57
Membranes were prepared exactly as Examples 47 to 50, e:xcept
as follo~vs:
Mixer rate was 1600 RPM
Casting resin solution temperature was 64C.
Table IX shows the product characteristics.,
Examples 56 and 57 are not within the scope of this in~entionj
they are included to illustrate the effect of using bath concentrations of
less than about 20~C formic acid.
This ~oup of Exa.mples also illustrates the advantages of baths
in the range neal to 46. 5~c in producing membranes with minimum pressure
drop at a given particle remo~al rating.
~5
o~
s~
~ a~
:: ~ ~
:~3 h h h ~ a~
O ,~,q o o P. s
.
~ .'
a~
^ OD 0
~ Q~ h O O O .~
~'
K ~1 o o o e~
¢l . . .~ ~ .
~ ~ ~ ~ N O Nd'~ p ~ ~
~ ~ U~ D 4~
. ,~
O C~ O ~ ~ Cl~
_.~ O C~
~1 _ ~C;l C;~ h
.~ ~r ~ ~ eD ~ ~ o~ C~ O
O :CI O
~ ~ .~ ~ N ~1 ~ o ~ ~
~3 ~~
.,. ~
~ - .
7~
~6~(~07
E~AMPLES 58 and 59
_ .
These membranes were prepa~ed using the same procedure as
E~amples 4~ to 50 except as follows:
~a) Starting resin concentration was 17~c.
(b) ~astin~ r esin solution was prepared from 3~4. 7 grams per minute
of star$ing resin solution using as nonsolvent diluent a solution containing 32. 8~c
formic acid in water, delivered to the mixer at a flow rate of 1~2.1 grams/
minute.
(c) Mixer speed was lgOb RPM.
(d) Composition of the ca~ting resin solution was: 12. l~c resin, and
67. 8'3~ of formic acid.
te) After filtering through a 10 ,uM filter, one-half of the so~ution
was further filtered through a filter with an approximate 0~ 05 to 0~10 ,uM
removal rating. The two portlons were then cast as films into a 46. 5% formic
acid bath, as Exarnples 58 (Eiltered 10 ,uM only) and 59 (~iltered lOIlM and
^~ O. 05 to 0O 10 IlM ) Data for the two, measured on a single thickness of
each, are listed in Table X.
- TABLE ~
Degree of filtration ~p, Estimated Absolute
20 ~3xa~ple of casting resin,~L5 inches t Tp~ particle
No. ,uM PSI ~f Hg mils ~?s.) rating, ,u~
. _ _
58 10 33.0 q.1 ~.~ 9 9x108 0~4
59 0 10 27.2 11.6 7.0 61 8XlOg û.65
The 11. 6 inch pressure drop of Example 59, resulting from the -ine
25 iltration step, should be compared with that of a normal product of this
invention with the same thickness and KL values, which would be appro~ima~ely
3. 5 inches of mercury.
77
O~
EXAMPLES 60 to 64
In these Examples, polyhexamethylene adipamide (Nylon 66) was formed
into membrane sheets usin~ a small batch proced lre. ~ 20% starting resin
solution was prepared by dissolving resin oE molecular weight =34000 in
5 ~8. 5% formic acidO A quantity of 500 grams of this solution was heated to
65C in a glass jacketed vessel approximately 4 inches inside diameter by
8 inches high, fitted with a two-inch diameter propellor-t~pe agitator, and
an externally operated flush val~e at its bottomD
- A nonsolvent solution was prepared containing 121 77% formic acid,
the balance being water. With the agitator rotating at 300 to 500 RPM, 241 g
of this nonsolvent solution was pumped into the apparatus, at a constant
rate, over a period of 2 minutes, the inlet nozzle being ~ mm inside diameter,
and located 1/4 inch from the arc described by the rotating propeller. During
the last portion of the two minute period, resin was seen to precipitate at the
15 inlet nozzle, all of which subsequently redissolved except for a small quantity
of lumps of resin about l/8 inch in diameter.
About 20 grams of the casting resin solution so formed was with~rawn
through the bottom valve, passed through a 42 mesh screen to remove lumps, and
without delay spread on a glass plate as a thin -Eilm, using a 0/ 010 inch doctor
20 blade, and the film then promptly immersed in a bath containing formic acid
and water, at 25C~
The membranes were allowed to set for several minutes, then were
stripped from the glass plate, washed in water and dried by exposure to
înfrared heat. The properties of the resulting membranes are shown in Table X1O
7~
~l~VC~07
O h
h tg S~ h h ,_,
P P P'
.
' C~ e~
~ ., O O O O
~ '
¢ ~L h v .~
~5 . ~3
~P e-l A X X X N
~:1 CQ ~ 0~ O-C~
E-~~ ._, .
C~ ~ O
~ di ~ ~ oo
.
~ ~ Q e~ î ~ ~ O
O~.S ~:
.
h ~ o O c~ O g
~ K c~
~v
o o ~ C~
79
~L6~(~07
Examples 60, 61 and 62 illustrate the effect o~ degree of nucleation
on product characteristics. Examples 60 and 61 are properly nucleated, and
~ield products with favorably low pressure drop, for their removal ratings.
In Exarnple 62, the higher rotation rate resulted in a casting solution with too5 low a degree of nucleation, and as a consequence, a relatively high pressure
drop.
,
C~7
EXAMPLES 65 to 68
In these Ega~ples, the polyamide resins shown in Table g~I
below were formed into membrane sheets using a small batch procedure.
A 20% starting resin solution was prepared hy dissolving resin of molecular
5 weight = 34000 in g8. 5% formic acid. A quantity of 500 grams of this
casting solution was held at ambient tempera~ re in a glass jacketed
vessel approximately 4 inches inside di~meter by 8 inches high, fitted
with a two-inch diameter propellor-type agitator, and an extelnally operated
flush valve at its bottom.
As the nonsolvent wa~er was used in Egamples 65 to 67, while in
Example 68 dimethylformamide was used. With the agitator rotatin~ a~
500 RPM, the nonsol~Tent was pumped into the apparatus a~ a
constant rate over a period of 2 minutes, the iT~et nozzle being 2 mm
inside diameter, and located 1/4 inch from the arc described ~y $he
15 rotating propeller. During the last portion of the two minute period, resin
was seen to precipitate at the i~et nozzle, all of which subsequently re-
dissolved e~cept for a small qua~tity of lumps of resin about 1j8 inch in
diameter.
A~bout 20 grams of the casting resin solution so formed was wîth-
20 drawn through the bottom valve7 passed through a 42 mesh screen to removelumps, and without delay spread on a glass plate as a thin film, using a
O. 010 inch doctor blade, and the film then promptly immersed in a bath
containing formic acid and wa;ter, at ambient temperature.
The membranes were allo~ed to set for several minutes, then
25 were stripped from the glass plate, washed in water and dried by ex3?osure
to i~rared heat. The properties of the resulting membranes are shown
in Table Xll.
81
6~3~30~
.~ .~ h .~
~) ~ ~ ~ h ~:
P P E~
CO
.
C~
~q
c~
~'~qL~ r-
~- ~ . ~
O . M ~1
d~
.~ ~ ~
O
U~
h ~ ~D ~
.
.
~ ~ tQ
~~ O ~ $ ~s O
9 ~ ,i, .
~ ~ ~
o
r- O
~ ~ a 8 o
.IZ >~,~ ~ R ~ h ~3 h
~ g 3 ~ 8 ~ a~ o ~ ~ ~ o ~ h
a~ s~
O IL~
CD C~Ct~ N~
82
