Note: Descriptions are shown in the official language in which they were submitted.
07 0005
~4~'7
This invention relates to hollow semi-permeable
fibers which are intended for use i~ fluid separations;
to aggregates (bundles) of the hollow fibers; and to
separation apparatus containing the aggregates of the hollow
Eibers.
By this i~vention there are provided hollow fibers
which are particularly advantageous for use in fluid ~eparation
apparatus in which the fluid separation is efected by
selective permeation through the hollow fibers. The hollow
fibers of this inventlon enable desirable membrane surface
area to be provided per unit ~olume of separation apparatus
and can provide enhanced separation efficiencies.
Many proposals have been made Eor the use of
semi-permeable membranes to eff~ct the separation by
selective permea~ion of at least one flu~d ~rom a fluid
mixture containing a~ least one other fluid. The commercial
adoption of semi-permeable membranes for 1uid separations,
however, has been limited, and presently larOe scale commercial
use o~ semi-permeable membranes iQ primarily only for
water desalination. ~hile semi-permeable mem~ra~es ha~e
been developed which exhiblt suitable select~vit~ of separation
or many operations, a fundamental difficul~y which has
been obser~ed is the relatively low flux which can be
obtained through the semi-permeable membranes, Accordingly,
large surface areas of the semi-permeable membranes must be
provided in order to obtain desirable quanti~ies of permeated
produc t .
The configuration of the semi-permeable membranes
significantly influences ~he Emount of active membra~e
surface area which can be o~tained in a given volume o~ a
semi-permeable membrane-containing separation apparatus. A
highly desirable membrane configuration Eor providing high
~ 3~ ~ 07-0005
ratios of active surface area per ~mit volume o~ separation
apparatus is a hollow fiber, or hol.low filament~ For
instance, Mahon in United States Patent No. 3~228,877 states
a~ column 2, lines 35 et. seq., ~hat with the semi-permeable
membranes in hollow fiber form ten thousand square feet or
more of active surface area can be provided per cubic ~oot
of volume of separa~ion apparatus. To accomplish such high
ratios Qf membrane surace area per unit volume of separation
apparatus, Mahon teaches that the hollow fibers should have
relatively small outside diameters and that the more
advantageous range of outside diameters of the hollow fibers
is between 10 to 15 microns. Others ha~e similarly found
that hollow fibers having relatively small outside diEmeters
are ad~antageous for utilization in separation apparatus
in order to provide suficiently large membrane surace
areas. For Example, Maxwell et al, disclose in United
States Patent No. 3,339,341 that hollow fibers ha~i~g outside
diæmeters between 20~and 250 microns are especially preferred.
In fact, the pate~tees state a~ column 14, line 63, that
pilot plant work was done utilizing hollow fibers having
ou~side diameters of 29.2 microns.
The use of relatively small diameter hollow fibers
, as semi-permeable membranes provides advantages in addition
to the ability to achieve high ratios of active surface
. 2'5 areas per unit volume of a separation apparatu~. As noted
-~ by Mahon at column 10, lines 57, ~ g., the amount oE
: pressure diferential which can be withstood by a hollow
fiber is directly related to ~he ratio of the thickness of
: the wall of the fiber to the inside diameter of the fiber.
Since in many separation opera~ions the higher the pressure
-2-
:
07-0005
`7
differential across the membrane the greater the flux which
can be obtained, the provision of hollow fibers which can
withstand high pressure differential is desirable. Mahon
thus concludes that the smaller the diameter of the fiber,
the smaller the corresponding wall ~hickness chat is necessary
to withstand a gi~en pressure drop, and walls of lesser
thicknesses advan~ageously exhibit less resistance to
permeate 10w; and hence provide greater fluxes, than
exhibited by walls of greater thicknesses.
Although previous proposals have been made to provide
relatively large active membrane surace area per unit volume
of separation apparatus, such apparatus may not peror~
adequately for a 1u~d separation in the environment in
which the fluid separation must be conducted for th~ apparatus
to be advantageous on economic and processing bases. For
instance, in assembling the hollow fibers in a separation
apparatus, several problems can occur which reduce ~he .
efectiveness of the separation apparatus. Firs~, hollow
fibers can, and essentially always do, contact other fibers
i~ the assembled separation apparatus. The resulting area
-- ~ of contact is unavailable ~o effect the desired separation,
and thus the ~lux and efficiency which can be obtained axe
reduced. Second, the con~act of the hollow fibers hinders
, .,;~
the ~low of fluid around and ~e~een the hollow ibers
thereby resulting in a non-uni~ormity of flow within the
. apparatus and even in localized pockets of fluid. These
pockets of fluid, when in contact with the e~teriors
of the fibers, contain an increased concentration of
a less permeable fluid of the feed. This greater
concer.tration of the less permeable fluid of the feed
~ 3
. ...~ ..-
.'~ . ' .
' '
_~ 07-0005
causes an increased permeation of the less permeabLe fluid
through the membranes and hence reduces the selectivity
of separation. In extreme cases iIl which the feed is a
liquid, the pockets of fluid may become so saturated wi~h
the less permeable fluid that the :Less permeab~e fluid
precipitates or separates between the hollow fibers.
Third, since the genexally hollow fibers have relatively
small outside diameters and thin walls, they are very
flexibLe. Thus, even if the hollow fibers are assembled
in the separation apparatu~ in a manner which minimi~es
the contact between the hollow fibers, the hollow fibers
may come i~ increased contact with one and another and
form uneven distribution channels due to the ease of
movement o~ the highl~ flexible hollow ~ibers during operation
of the separation apparatus.
Another important factor in~olved in the consideration
of whether or not a separation apparatus is advantageous
on economic and processing bases is the effect o~ the
separation apparatus on the energy of the fluid stream being
processed. Proposals have been made ~o empLoy semi-permeable
membrane-containing separation apparatus to selectively
separate one or more fluids from a fluid mixture con~aining
at lea~t one additional fluid wherein the fluld mixture
(retentate) is subje~tet to processing subseque~t to the
- 25 separa~ion operatian. If the separa~ion apparatus provides
significant resistance to the flow of the fluid mixture,
substantial energy expenditures may be required to
recompress the fluid mixture (retenta~e) to d~sired
pressures for the subsequent processing. The pressure drops
to the fluid mixture caused by the fluid flow resistance
,.
' '
.
~ 4
. , ,- ;., ,
, s - . ~. . .
~43~ 7 Q7-0005
of the sep~ration apparatus are frequently subs~antial
when the fluid mixture is fed to the bores of the ~ibers.
For example, Gardner, et al, disclose, inter alia, in
L~G~ .3~Y~a~C~ - L3~ - C~ Oct~ber 1977, pages 76 to 78,
the usa of hollow fiber membrane-cantaining separation
apparatus for removing hydrogen ~rom a hydrogen and carbon
monoxide feed stream in an Oxo alcohol synthesis plant.
The feed str~-~am~ is at a pressure of 350 pounds per square
inch gauge (psig), compressed to a pressure o~ 600 psig,
passed through the bores of the hollow fibers in the
separation apparatus, a~d recovered as a retentate stre&m
rom the separation apparatus a~ 330 psig. Clearly, the
compression of the feed stre~-lm is expenslve :in terms of the
capital expenditure for the compressor and the operating
costs fcr the compressor. Although a substantial pressure
drop is i~curred due to the bore-side feed of thé ~eed itream
to the separation apparatus, the bore-~ide ~eed is apparently
necessarv due to the lack of distribution and loss of
e~iciency i~ the feed stream~ were fed to the exterior
(shell) side of the hollow fibers.
S~ell side feed to hollow fiber-containing separation
apparatus can pro~ide oth~r ad~antages. For in3tance, a
greater surface area for ef~e~ting the separatio~ is provided
~ ~ at the exter1Or surface~ of the hollsw fibers than at ~he
j 25 interior surfaces of the hollow fibers. Moreover, hollow
fibers may be able to withstand higher pressure differ~entials
when the higher pressure is at the exterior as opposed to the
interior of the fibers c~lnce generally materials exhibit
greater compressive than tensile strengths.
,: ~
~ 5-
, .: .
.~ ,;. ..
~j 07-0005
Effor~s have been expended to provlde hollow fiber
~embrane-containing separation appara~us having improved
fluid distribution between the hollow Ei.bers. Rosenblatt
in United States Patent No. 3,616,92~ di.scloses the use of
highly crimped hollow fibers for use as the semi-permeable
membranes. The crimped hollow fibers are adhesively bonded
to one and another at a plurality of the abu~ing areas in
order to maintain the spatial relationship between the hollow
fi~ers. One disclosed separation apparatus is prov:ided with
means to introduce the feed at the periphery of the assembly of
hollow fibers such that the feed flows radial~y inward
through the hollow fibers. The patentee provides no general
irldication of the proportion of the cross-sectional area
of the se~aration apparatu~ which is occupied by the hollow
fibers (i~.e., packing factor or packing density), however,
this proportion appears to be relatively low~ e.g., about
16 percent in Example 4, as compared to con~entional
s~paration apparatus in which the feed is introduced into ~ -
the bores of the hollow filaments (of~en abou~ 45 to 60 or
more percent as illustrated by Maæwell, et al, in United
States Patent No. 3,339,341 at column 5, lines 10 to ~5).
The use of the low packing fac~ors as appa~ently suggested
; by Rosenblatt is directly contrary to the desire for
minimizing the size of the separa~ion apparatus, Moreover,
the essential adhesive bonding of ~he hollow fibers to
maintain their spa~ial relatio~shlps requires an additional
processing step, and the presence of ~he adhesive rPduces
the available membrane surface area for effe~ting the fluid
separation.
In accordan~e with this invention hollow, semi-
permeable fiber~ which are intended for use in fluid
..
.,
-6-
- 07-0005
'7~
~epara~ions have relatively low amplitude waves, or crimps.
These hollow fibers are particularly advantageous for
assembling bundles of hollow fiber~; which are oriented
substantially parallel to one and another. The bundles can
S be assembled to provide desirably high packing factors to
enable separation appara~us having a large membrane surface
area to be advantageously compact ln volume, Moreover, even
though high packing ~actors are provid d, good f}uid distri-
bu~ion throughout the bundle can be obtained. Thus the fluid
mixture containing at least one fluid to be permeated can
be fed to exteriors of the hollow fibers with the permeate
be~ng facilely remo~ed from the interiors of the hollow
fibers. Moreover, assembly o~ the hollow fibers to form a
b1mdle can be non-complex and not re~uire spacing means or
special techniques to provide a packing factor which permits
fluid dispersion through the bundle. Advantageously, the:
hollow fibers in the bundle need not be fixed in a relationship
to other hollow fibers by the use of an adhesive ma~erial
in order to maintain a de~irable packing factor thro~ghout
- 20 the bundle when employed in a fluid separation operation.
sir.ce good fluid dispersion can be obtained through
~: :. bundles compri~ed of hollow fibers of this invention, a
~: fluid mixture containing at least one fluid can be permeated
through the membrane (permeat~ng fluid~ ca~ be introduced to
~ 25 the exterior sides of the fibers (i.e., ~he shell-side of the
:- ~ bundle), and attractive 1uid sepaxation efficienc1es ca~ be
~ obtained. Accordingl~, the fluid mixture can be processed
-~ ~ in and recovered frcm the shell-side of a fluid separation
pparatus containing a bundle of hollow iber membranes in
- : 30 aecordance with this invention, and the recovered fluid
mixture may be at substantially the same psessure as the 1uid
; ,;~ ' ~,
~ 7-
07-0005
~ 3'~
mixture introduced into the separatlon apparatus. Therefore,
expensive fluid compression apparatus may not be required to
recompress the recovered fluid ~o su~ioient pressures for
subsequent processing, and7 even if recompression is reiquired,
the size of the compression apparatus and the amount of
compression required may be significantly less than the size
of the appara~us and amount of compression required if the
fluid mixture were fed to the bore side of the bundle.
The advantages provided by the hollow fiber membranes
having the low amplitude crimp of this i~vention can
particularly be observed when the fluid mixture i~s fed to
the shell-side o the bundle as compared to when the fluid
mi~ture i~ fed to the shell-side of a bundle of hollow fibers
which have an essential absence of cri~ps~ These advantages
can be observed when ~he fluid mixture is fed radially, i.e.,
the fluid mi~ture is introduced in a mid-por~ion of the bundle
and 10ws substantiall~ perpendicularly to the orienta~ion
of the hollow ibers, or predominan~ly axially, i.e., the
fluid mixture is introdueed at a~ outside portion of ~he
~0 bundle, flows generally in the same tirection as the
orienta~ion of the fibers, and exi~s at another portion of the
bundle. While often radial feed i9 consldered to pr~vide
bet~er fluid separation efficieneies, advantRgeous ~luid
; separation efficiencies can be obtained employing axial feed
- 25 to bundles comprising the hollow fiber m~mbranes of this
~- invention. Axial feed may ~e desirable since the separation
appara~us may be less complex in design than radially-fed
separation apparatus, and sinoe no radial feed conduit need
be positioned within the bundle, ~he bundles employed for
axial flow may comprise a greater ratio available memibrane
. .,
~ surface area per given volume of separation apparatus than
.... " , ~;
the ratio of available membrane surace area per given ~o~ume
of radially-fsd separatlo~ appara~us. While shel.l-side feed
-8-
~: 07-0005
to the bundle is generally desirable, there may be separation
operations in which bore-side feed may be dasirable. For
instance, when a fluid mixture for processing by membrane
separation need not be main~ained at high pre~sures for
furth~r processing bore-feed may be attractive to recover ~he
permeated fluid at the shell-side with little pressure drop
subsequent to the permeation.
The hollow, semi-per~eable fibers o~ this invention
ha~e crimps, or waves, of low amplitude. Surprisingly, it
has been found that desirable dispersion o~ fluid mixtures
~hrough bundles of the hollow ~ibers can be obtained eve~
though the bundles have a relatively high packlng factor.
The amplitude of the crimps is no~ so large that an undue
amount of cross-over occurs when the hollow fibers are
assembled in a close-packed sub~9tantially parallelly-oriented
fiber bundle. The term "cross-over" as employed herein
refers to situations in which the crimp has such a large
amplitude that th~ crimp protrudes sufficiently far from the
axis of the hollow fiber that it bec~es positioned between
~wo or more adiacent hollow fibers and separates the adjacent
hollow fibers by at least the diameter of the fiber. Such
.
cross-overs tend to prevent ~he obtaining of bundles having
desirably high pac~ing factors.
:~ The amplituda of the crimp~ as referred to herei~
: 25 is one-hal the lateral distance betwe~n the midpoint o~ the
hollow fiber at one apex ~o the midpoint of the hollow fiber
at the next adjacent9 diametrically-opposed ape~. When no
: adjacent, diametrically-opposed apex e~ists, the æmplitude is
.
the lateral distance between the midpoint of the hollow fiber
` 30 at the apex to the midpoint of the hollow fiber which is not
;.~.. ~. crimped. Ad~antage~usly, the amplitude of ~he crimps is l~ss
~ ' _g_
~ .
07-0005
~ `7
than about S0 percen~ of the diameter of the hollow fiber,
and generally, the amplitudes of the cr~mps are within the
range o about 1 to 30 percent of the diameter of the hollow
fiber. Fiber crimp amplitudes of above about 50 percent of
S the diameter can also be employed; however, g~nerally after
the bundle of fibers is assembled, the bundle must be
compressed to obtain a desirably high packi~g factor. Some
compression of the bundle serves to maintain the hollow fibers
in a substantially fixed relativnship to each other, and thus
the tendency of the hollow fibers to move such that lateral
flow channels, which flow channels decrease the efficiency o
separation, is generally abated. The c~mpressio~ should not
unduly affect localized regions o~ hollow fi~ers 9UC~ that
fluid penetration is inhibited in those region~ or that such
non-uniform loading of the hollow ~ibers is provided that
the hollow fibers may collapse. Each of the crim~s in a
hollo~ fiber or 2mong the hollow fibers employed to fo~m
the bundle may have the same or diferent amplitude than
; other crimps, and the ~mplitudes o the crimps may vary over
a range to assist in breaking any register between hollow
fibers. Moreover, the bundle may conta.in hollow fibers having
substantially no crimps, which fibers are i~terspersed with
hollow fibers ha~in~ crlmps. For example, hollow fib~r~
ha~ing a distribution of crimp amplitudes of from about 10
to about 30 percent of the diameter of the hollow fiber have
been assembled into a bundle having a packing factor of abo~t
50 percent which bundle exhibits good fluid dispersion when
axially fed.
~- Hollow fiber diameters may be selected over a wide
range, however7 th~ hollow fiber should have sufficient
wall thickness such that the crimp is maintained. Frequently,
~` the outside diame~er of the hollow fibers is a~ least about
. -. -;:....
-10-
07-OOOS
'~L$~L4L3f~7
50, say, at least about lOO,microns, and the same or different
outside diameter fibers may be contained in ~ bundle. Often,
the outside diameters are up to about 800 or 1000 microns.
Although larger outside diameter hollow fibers can be
S employed, they are less preferred due to the low ratios of
hollow fiber surface area per unit volume of fluid separation
apparatus which are provided. Preferably, the outside diameter
of the hollow fibers is about lS0 or 350 to 800 microns.
Thus, the amplitude of the crimps is often in the range of
about from 10 to 400 microns, say, about 10 to 300 micro~s,
with an average crimp amplitu~e of about 15 to 250 microns.
It has been found that the crimps need not be
conti~uous o~er the length of a ho~low ~iber in order to
provlde desirable hollow iber membranes for assembly into
a ~undle. Thus, ~he crimps may be intermittently spaced
ovPr the leng~h of the hollow fiber, and the freque~cy of
~he crimps may be irregular. Moreover, as stated above,
fibers with a distribution of crimp frequency ca~ be emp1oyed.
Ge~erally, at least about 50 percent, preferably at least
about 75 percent, of the fibers in a bundle are cr~mped. The
hollow fibers which are crimped frequently have an average
of at least o~e crimp per each five ce~timeters of fiber
length. The average frequency of crimps over the length
of a hollow fiber is often about 0~2 to 10 or more, say,
about 0.25 ~o 5, per centimeter.
If the frequency of the crimps in the hollow fiber
i irregular, the crimps generally range in frequency from
about 1 to about 50 crLmps per five centimeters, e.g., from
about I ~o about 30 crimps per five centimeters, of hollow
fiber length.
The period of the crimps, i.e., ~he lellgth of each
crimp, is desirably sufficiently short that the crimp maintains
.. ~
.....
its conflguration and substantial changes in amplitude oE
the crimp does not occur when the hollow fibers are assembled
into a bundle. For instance, if the period of the crimp
is too long and gradually ascends to its apex, then even minor
mechanical forces may tend to straighten out the hollow fiber.
In order to obtain the advantages provided by this invention,
the period of the crimp can be relatively short, e.g., less
than about 5 cen~imeters. The shortness of the crimps is
generally llmited by the dimensions of the hollow fiber, i.e.,
with smaller diameter hollow fibers generally smaller crimp
periods can ~e obtained. Frequently, the average crimp period
is about 0.05 to 5, e.g., about 0.1 to 2 centimeters. The
ratio of the average crimp period to the average frequency of
crimps may vary widely, for instance, from about 0.05:1 to
1:1, often about 0.1:1 to 1:1.
The amplitude, frequency of crimps, and crimp period
are factors which relate to the configuration of the hollow
fibers. A useful aid which encompasses these factors for
describing the configuration of the hollow fibers i5 the
ratio of the actual length of the crimped hollow fibers to
the length of the hollow fibers if they were straightened.
Op~ical analytical tools are available for such determinations,
such as image analyzers available from Quantimet, Monsey, N.Y.,
which do not require physical straightening of the hollow fibers.
In view of the small differen~es in crimped and uncrimped length, ~ -
a convenient procedure is to report the differences in percent
of length change due to crimping. The percent of length change
is frequently in the range of about 0.01 to 10, e.g~, about
0.05 to 5.
The hollow fibers of this invention can be assembled to
form bundles of any suitable configuration. Advantageously,
the hollow fibers are substantially parallel:Ly oriented.
The cross-section of the bundle may be any suitable shape
-12-
.4L3'r~7' 07-0005
for use in fluid separation apparatus, e.g., circular,
oval, etc. The packing factor of the bundle is influenced
by ~he ampli~ude of the crimps, the frequency o the crimps,
the period of the ~rimps, and the compression o~ the bundle.
Generally, the packing factor of the bundle is at least about
40 percent and may be up to 65 peroent or more. Often, the
packing factor of the bundle is about 45 to 6S percent. For
axially-fed separation apparatus, ~he packi~g factor of the
bu~dle is frequently about 45 to 55 percent. S~l~e the
packing factor can be maintained due to the con~iguration of
the hollow fibers, spacing means to provide pack:ing factors
in the desired range need not be employed. For bundles
havlng substantially circular cross-sectio~s, the diameter~
of the bundles may ~Jary widely, e.g., oten ~rom at Least
about 0.02 up to 1 or more me~ers. ~he~ the separation
apparatus is radially ~ed, the diameter o~ the bundle may
be greater than 1 meter with ade~uate fluid dispersion
throughout the bundle without ~esulting in undue pressure
drops. On the other hand, when the separation apparatus is
axially fed, it has been found that enhanced fluid dispersions
through the bundle are obtained with higher space veloci~ies.
Accordingly, sm~tller bundle diameters are often preferred,
e.g., about 0.02 or a . os to 0.5 meter. The efecti~e length
-- o~ the hollo~ fibers in ~he bundle may also vary widely,
for inst nce, from about 0.2 to 15 or 20 meters, e.g., about
1 to 10 meters,
The bundle may be encased (or potted) proximate
at least one end to prevent fluid commu~ication between the
e~teriors and interiors of the hollow fibers except through
the semi-permeable walls of the fibers. Any suitable
;~ ~; method for embeddi~g the fibers in the potting material can
be employed, e.g.~ by casting a pot~ing material around
-13-
.
the end of the bundle such as disclosed in United States
Patent Nos. 3,339,341 (Maxwell et all ancl 3,442,389 (McLain)
or by impregnating the ends of the fîbers with potting
material while assembling the hollow fibers to form the
bundle such as disclosed in United States Patent Nos.
3,45$,460 (Mahon) and 3,69 a, 465 (McGinnis et al~
In assembling the bundles/ it is desired that the
crimps in the hollow fibers do not fall in register. The
substantial avoidance of register can be achieved in
various manners. For instance, the fibers can be aligned
such that the crimps, e.g., of regularly crimped hollow ~ ~
fibers do not match. This procedure may be unduly complex. ~ ~;
Advantageously, at least some of the hollow fibers vary in
at least one of crimp frequencies, crimp period, and crimp `
amplitudes such that with a random assembly of the hollow
fibers, the probability of obtaining an undue amount of
fibers in register is minimal.
The hollow fibers may be fabricated from any suitable
synthetic or natural materîal suitable for fluid separations
2Q or as supports for materials which effect the fluid
separations. The selection of the material for the holiow
fiber may be based on the ~est resistance, chemical resistance~
and/or mechanical strength of the hollow fiber as well as
other factors dictated by the intended fluid separation
in which it will be used and the operating conditions
to which it will be subjected. In order to maintain
the de6ired fiher crimp of this invention, the fiber
should exhibit appropriate mechanical properties such that
-the crimps do not unduly dissipate with time or during the
separation operation. With hollow fibers fabricated from
materials having lesser strengths, it may be necessaxy to
-14-
07-0005
employ larger fib~r diameters and wall thicknesqes to
impar~ sufficient strength to the hollow fiber crimps that
they substantially retain their configurations. Often,
the wall thickness of the hollow fi.bers is at least about
5 microns, and in so~e hollow fibers, the wall thickness
may be up to about 200 or 300 microns, say, about 50 to 200
microns. In many instances the mat:erial of the llollow fiber
exhibits a relatively high tensile modulus, i.e,, modulus of
elasticity or Young's modulus, that the crimps can be
retained even under longitudinal and lateral stressing.
Often the tensile modulus (ASTM D638) is at least about
15 kilograms per square millimeter (~g/mm2)l e.g., at least
about 40 kg/mm2, and for some metals and allo~s, the te~sile
modulus is up to about 3000 or more kg/mm2. Most frequently,
pol~meric materials which are to be employed are selected
from those poly~ers which exhibit a tensile modulus of
about 60 ~o SOO kg/mm2.
In order to provide desirable fluxes through the
hollow fibers, particularly using those hollow fibers having
walls at least about 50 microns in thick~ess, the hollow
fibers may have a substantial void volume. Voids are
regions within the walls of the hollow fibers which are
vacant of the material of the hollow fibers, Thus, when
voids are present, the density of the hollow fibar is less
than the density of the bulk material of the hollow fiber.
-- Often, when voids are desired, thç void volu~e of the holl.ow
fibers is up to about 90, say, about 10 to 80, and sometimes
about 20 or 30 to 70, percent based on the superficial
volume, i.e., the volume contained within the gross dimPnsions,
of ~he hollow fibers. The density of the hollow fiber can
be essentially the ~ame throughout its wall thiekness, i.e.,
-15-
':
07-0005
3~
isotropic, or the hollow fiber can be charac~erized by having
at least one relatively den~e region within its wall thickness
in barrier relationship to fluid flow through the wall of
the hollow fiber, i.e., the hollow iber is anisotropic.
Generally, a relatively dense region of anisotropic hollow
~ibers is substantially at the exterior of the hollow fiber.
The material for ~orming the hollow fibers may be
inorganic, organic or mixed inorganic and organic. Typical
~ inorganic materials include glasses, ceramics, cermets, metals
and the like. The organic materials are ~sually polymers.
In the case of polymers, both addition and condensation pol~mer~
which can be fabricated in any suitable manner to provide
hollow fibers are included. Gensrally organic and some times
organic polymers mixed with inorganics (e.g., filler~) are
used to prepare the hollow fibers. Typical polymers can be
substltuted or unsubstitu~ed polymers and may be selected
from polysulfones; poly(styrenes), i~cluding styrene-containing
copolymers such as acrylonitrile-styrene copoly~ers, styrene-
butadiene copolymers and styrene-vinylbenzylhalide copolymers;
polycarbonates; cellulosic poly~ers, such as cellulose
acetate; cellulose-acetate-bu~yrate, cellulose propionate,
ethyl cellulose, methyl cellulose, nitrocellulose, etc.;
polyamides and polyimideq, including aryl polyamides a~d
..,, .,.~ .
aryl polyimides; polyet~ers; poly(arylene oxides) such as
poly(phenylene oxide) and poly(xylylene oxlde); poly(ester~
amide-diisocyanate); polyurethanes; polyesters (includi~g
polyary}ate~), such as poly(ethylene lerephthalate),
poly(alkyl methacrylates), poly(alkyl acrylates), poly
~phenylene terephthalate~, etc.; polysulfides; polymers
from monomers having alpharolefinic unsaturation other than
. ~ mentioned above such as poly(ethylene), poly(propylene),
poly(butene-l), poly(4-methyl pentene-l), poly~inyls, e.g.,
paly(vinyl chloride), paly(vinyl fluoride), poly(vinylide~e
16
07-0005
chloride), poly(vinylidene fluoride), poly(vinyl alcohol),
poly(vinyl esters) such as poly(vinyl ace~ate) and poly(vinyl
propionate), poly(vinyl pyridines), poly(vinyl pyrrolidones),
poly(vinyl ethers), poly(vinyl ketones), poly(vinyl aldehydes)
such as poly(vinyl formal) and poly(vinyl butyral)) poly
(vi~yl amides), poly(vinyl amines), poly(vinyl urethanes),
poly(vinyl ureas), poly(vinyl phosphates), and poly(vinyl
sulfates); polyallyls; poly(benzobenz~midazcle3; polyhydrazides;
polyoxa~iazoles; polytriazoles; poly(benzimidazole)i
polycarbotiimides; polyphosphazinesi etc., and interpolymers,
including bloc~ terpolymers containlng repeating units from
the above such as terpolymers o~ acrylonitrile-vinyl bromide-
sodium salt of para-sulfophenylmethallyl ethers; and grats
and blends contai~ing any of the foregoing. Typical substituents
providing substituted polymers include halogens such as
fluorine, chlorine and bromine; hydroxyl groups; lower al~yl
groups; lower alkoxy groups; monocyclic aryl; lower acyl
groups and the like.
The crimps m.ay be i~duced into the ~ollow fîbers i~
any suitable manner. For instance, straight hollow fibers
. may be sof~ened with solvent or plas~iciæer for the material
o the hollow fiber, mechanically deformed to ~mpart the
crimp configurat~on, and then treated, e.g., by drying, to
~ r~move the solvent or plasticizer such that the hollow fiber
: 25 regai~s the desired -.oigidity. Alternatively, or in addition,
the ~aterial of the hollow fiber may be softened by the
application of heat to the hollow fiber. In any event, the
softening is sufficient such that the bore remains
substantially u~restricted upon application o mechanical
force to provide the crimp. A co~venient method for
:
, - .
~ 17-
, " ~j,~ , .
~7-0005
3~ 7
providing crimps in coagulation spun hollow fibers, i.e.,
hollow fibers which are spun from a solvent solu~ion of the
material iIltO a non-solvent for the material, is by winding
the spun hollow fiber onto a bobbi,n while wet. Upon loss o
solvent, and drying if the hollow fiber is dried on the bobbin,
hollow fibers tend to shrink and ehus put an increased
pressure on the underlying fibers. This pressure provides
the mechanical force required to impart the desired crimps,
and the solvent loss enabLes the fibers to gain additional
rigidity such that the crimp is retained. Slnce the forces
exerted on the hollow fibers often vary with the depth of
the hollow fiber within the bundle, irregular crimp patterns
occur with the hollow fibers from the outer portions of the
bundle tending to have fewer, ~ore widely-spaced crlmps
than the hollow fibers closer to the center of the bundle.
The following examples are provided ~o ~urther
illustrate the invention. All parts and perce~tages of
liquids and solids are by weight, and all parts and
percentages of gases are by volume, unless otherwise
indicated.
;,. .~
EXA~E 1
A hollow iber is preparet from dried polysulfone
polymer having the repeating unit
CX3 0
.
~ ~18~
., . ~ .
. . .
. . ..:,
07-OOOS
where n, representlng the degree of polymerization, is
about 50 to 80 and is available from Union Carbide under
the designation P-3500. The polysulfone is admixed with
dimethylacetamide to provide a dope containing about 27.5
weigh~ percent polymer, and the dop~ is coagulation spun
in~o water at a te~perature of about 4C through a
spinnerette which is immersed in the water. Thç spinnerette
has an outer orifice diameter af 0.0559 centimeters, an
inner pin of 0.0229 centime~ers, and an injection port
of 0.0127 centi~eters through which water is in~roduced.
The dope is pumped and metered to the spinnerette at a rate
of about 7.2 milliliters per minute and is drawn from the
spinnerette a~ a hollow fibe~ at a rate of about 33 meters
per minute. Ater the coagulation has substantially
occurred, the hollow fiber ~s washed in room temperature.
The hollow fiber is wound substantially without tension
on a 12 inch (approximately 25.4 centimeter~ between inside
heads) bobbin with a bobbin winder, i.e., the hollow fiber i5
fed through an axially traversing guide (which reverses at
each end of the bobbin) and is collec~ed on the surface of
a rotati~g bobbin so that the hollow fiber is wound on the
bobbin in sequential layers of helical coils. The bobbin
i stored in an aqueous vat at room temperature during
which time the fibers on the bundle shrink to impart crimps.
The hollow fi~ers are then wound on a skeiner having about
a six meter circumerence. The hollow fibers are removed
as three meter long hanks and are hung ~nd allowed to dry
at ambient labora~ory temperature and humidity. The hollow
fibers hzve an outside diameter of abou~ 540 microns a~d an
- 30 inside tiameter o about 260 microns.
A random sample of hollow fibers are removed from the
dry hanks and are analyzed for configuration charac-
,
I teristics on an image analyzer obtained from Quantimet
.. i .
. . -, , .
-19-
of Mon~ey, New York. The random sample contained samples
of fibers which were obtained at the inside, middle and
outside portions of the bobbin. The results (approximate
frequency oE occurrence) are presented in Table I.
EXAMPLE 2
Crimped, hollow fibers prepared by substantially
the process set forth in Example 1 are assembled in the
form of a bundle of substantially parallell~oriented hollow
fibers. Approximately 1200 hollow fibers having a length
of about 30 centimeters are employed, and the fibers are
randomly selected from fibers obtained from all portions
of the bobbin. The hollow fibers form a cylindrical bundle
which is about 2.5 centimeters in diameter (about a 50
percent packing factor). Epoxy seals are fabricated at
both ends of the bundle by sealing the ends of the bundles
and then immersing the ends (tube sheet and plu~ ends) in
~iquid epoxy resin and allowing the epoxy to cure. After
curing, a knife is used to open the bores of the hollow
fibers at the tube sheet end.
The bundle is immersed in a solution of 5 weight
percent of the product marketed under the trademark SYLGARD
184 in isopentane. SYLGARD 184 is a cross-linkable dimethyl-
siloxane polymer which is availahle from Dow Corning
and cures at ambient temperatures. The bores of the
hollow fibers are in communication with a vacuum of
about 600 to 700 millimeters of mercury. The immersion
is for about 15 minutes, and the vacuum is con-tinued ;
for about another 15 minutes after the bundle is emerged
.: . ~
-20-
~ .
, . . . .
3~
07 -0005
_ ~ __.
~ ~ ~ C~l o
~ --I
~ c), ~ ~r ~
' ~ --I - .
~, ~ o ~
~ _l ,_ _ .
., o o ~ ~
_ a _ _
o
~ o
~: ~! o '5'
. : o
.. o ~ ~o o
o _~
.: . ~ o C~ ~ o
i~ e ~
,. e~ o ~ ~ o
. h _I
~ ~ ' ; ; l u T .
''.~:;.,. ~ O ~ ~ ~ ~
0 3 ~::
. ~O~~~ e _
~ ~ I
,:..;.
3'~7
from the SYLGARD* 184 solution. The coated bundle is cured
at about 4 n to 50C for about 24 hours and installed in
an axially-fed fluid separation apparatus. A gaseous
feed comprising hydrogen and carbon monoxide is introduced
into the shell side of the separation apparatus and the
permeabilities of the permeate gases are determined. The
permeabilities are determined using the log mean partial
pressure drop across the hollow fibers. A separation
factor is determined by dividing the permeability of hydrogen
by the permeability of carbon monoxide. A separation
efficiency is also determined b~ dividing the separation
factor calculated for the gaseous feed comprising hydrogen
and carbon monoxide by a reference separation factor deter-
mined from the separation factor determined from the separate ;~
determination of the permeabilities of essentially pure
hydrogen and essentiall~ pure carbon monoxide. ~ -
Lower separation efficiencies are often indicativa
of poor fluid dispersion in the bundle such that localized
areas of high concentrations of the undesired component
~carbon monoxide) occur and thus increase the permeation
of the undesired component and lower the separation factor.
The results are provided in Table II. ;
, ~ .
'~ ': '
* Trademark ~ ~
; ~ ' '' ' ,
~ ~ -22-
:':
' .' .
3~t7 07-0005
~ . . _
o
C
~, o ~ C~ ~ ,, C~ ` CO ~ oo
U Cr~ ,` ~o ~ ~ ,~ ~o ~o ~ ~ ~ ~
U~ ~
_I_ _ ~_
o ~ Y ~ ~ ~ ~.
o
~ U _ , _ __
C~. ~ ~ ~ e~ ~ ~ ~ ~ e~l ~'`I ~`I ~ ~ C`l
U~ ~
~3--~ ~_ __
~U
~ U~
u O ~ C~ : 00 ~ ~r ~ ~ co ~1
~D
U o ,, ~ ~ ~ ~ ,, _, ~ ~ ~ ~ ,~
~X
~ ~ _ _ .
`O `D u~ ~ ~;r ~ cr. u~ ~r
= ~
~ .
~ .
_ .
~ , :
~ O ~ I~ r~ r~ I~ c~ ~o r~ ~ C)
1~ i~ C~ ~ C~
O ~ ~
:~ ~O~ ~1 ~ I
_
Cr'
~ Ir~ O ~d' O ~O ~'7
_~_ l
::. . r~:1 t~
R u r~ 0 ~ 1~ o ~ 1
~1 ~ ~
l~i __ _ ,._~_
â ~ 0~ 0~ ~
U~ ~0 ~ C~ ~ ~ ~ ~ ~ ~ ~ C~
. ~ _ _
3 _~ c~ o ~ `D
o a~ ~ cr~ t CO ~ I ~ O
~:'' ~ U~ ~ 1
_
a
~, ~o ~ ~0 c~
~ ,~ e2 ~ 1~ r~. ~ ~ ~ ~ ~ ~ ~ ~ o
~ o . ,_ ,_ _ _ _
~I; -~ ~ 0
oo _i ~1 ~ ~ co ~0 CO co ct~ oO r~
. ~ a ~ ~ ~ ~ ~ ~ ~ ~ c~
.. ~ ~ V~
__ _
-. o~ ~
.,. ~ .
23-
. . ~, ... ; .
~ 3~7 07-0005
E.YAMPLE 3 (Co~parative)
The procedure of Example 2 is essentially repea~ed
except that the hollow fibers employed are not wound on
bobbins af~er spinning and do not have crimps. Since the
hollow ibers have an essentiAl absence of crimps, the
bundle tends to be smaller in diameter. In the runs, the
gaseous feed comprise.~ 22 volume percent hydrogeIl and 78
volume percent carbon monoxide. I~ ~wo runs, ~he shell
pressure is about 2~ ahmospheres absolute and the bore
pressure is about 4.1 atmospheres absolu~e. At a eed rate
of 12.3 liters per minute (STP) the hydrogen permeability
i5 observed to be 37.7 x 10 6 cc/cm2-sec-cmHg, and the
carbon monoxide permeability ls 1.60 x 10 6 cc/cm2-sec-~mHg.
The separation factor is calculated to be 23.6 which is
at a separation efficiency of about 63 percent. At a higher
~eed rate, 21.2 liters per minute (STP), the hydrogen
permeability is 41.4 x 10 6 cc/cm2-sec-cmHg, a~d the carbon ~:
monoxide permeability is 1.74 x 10 6 cc/cm2-sec-cmHg. The
separation factor i~ calculated to be about 23.8 which is
at a separation efficiency of 64 percent. In both of these
runs, calculations based o~ flow rates and permeabilities
only accounted for less than 95 percent of the hydxogen
introduced to the separation apparatus in the feed. The
procedure is again repeated at a shell pressure of 2~
atmospheres absolute, a bore pressure of 2.5 atmospheres
absolute, and a feed rate of about 14.5 li~ers (STP) per
minute of a blend gas consisting of 27.2 percent h~drogen
and 72.8 percent carbon monoxide. The hydrogen permeability
i~ about 31.6 x 10-6 cc/cm2-seo-cmHg, the carbon monoxide
permeability i9 about 2.0 x 10 6 cc/cm2-sec- ~g, and the
separa~io~ factor is 15 . 8 . The reference separation factor
-24
~, : ....... .
~ 7 07-0005
is determined to be 36. a to provide a separation efficiency
of 42. Virtually all the hydrogen is accounted for by
material balances.
Since the diameters of the bundles tested in
S Exæmple 2 and 3 are relatively small, the differences in
separation efficiencie~ are not as pronounced as they might
be in a comparison of larger bundles in which greater
penetration into the bundle is required,
:
' . .,
25-
.:
~ . .
