Document Eqn6LVG7JdqjM1kVD3pjBYJ5V

m Advance Copy ( pall) QUALITY CONTROL OF ABSOLUTE BACTERIA REMOVAL FILTERS INTRODUCTION: A chievem ent of b acterio lo g ical ste rility re q u ire s that: A. The filte r m u st reliab ly accom plish its rem oval function. B. The filte r m u st not re le a se into the dow nstream fluid significant quantities of foreign m a te ria ls - e ith er in solution or as solid p articles. C. The filte r is p re fe rre d to have long life , low p r e s s u r e d ro p , ease of in stallatio n and sterilizatio n , freedom fro m problem s such as eiir binding, and low c o st. We shall co n cen trate in this p ap er on the m eans fo r a ssu rin g co n sisten t quantitative rem oval of b a c te ria follow ed by a b rie f rev iew of oth er quality control facto rs. SIMULATED SERVICE T E S T : D uration of filtra tio n o p eratio n s v a rie s fro m a few m in u tes up to s e v e ra l m onths o r y e a rs of continuous o r in te rm itte n t flow . In o rd e r to p rovide the h ig h e st p o ssib le re lia b ility , w hile b earin g in m ind that a la rg e num ber of te s ts m u st be run w ithin a reaso n ab le length of tim e, P a ll's b asic sim u lated serv ice te st has been se t up as follow s: A suspension of the sm allest co m m ercially available b acteriu m , Pseudom onas d im in u ta (ATCC 19146), in w a te r a t 2 to 4x10^ p e r lite r , is fed contin u o u sly at a ra te of one lite r/m in u te to the te s t c a rtrid g e o v er a p e rio d of 2 to 4 w eeks, by w hich tim e the elem en t m ay be clogged w ith b a c te ria to the extent th a t the one lite r flow ra te can not be m aintained. Sam ples a re taken daily and cu ltu red ; effluent m ust be consistently sterile. To supplem ent the above stan d ard te st, a sm a lle r n um ber have been run using h ig h er b a c te ria loadings (hence le ss tim e to clogging and sh o rte r te st duration), and others have been run using low er b a cte ria loadings, w ith longer operating cycles. NON-DESTRUCTIVE QUALITY CONTROL TEST: The developm ent of a sa tisfa c to ry n o n -d estru ctiv e te s t w hich c o rre la te s w ell w ith the above d escrib ed sim u lated se rv ic e te st has occupied m uch of o u r e n e rg ie s d u rin g th e la s t 18 m o n th s. C o n sid eratio n w as given to the u se of the " f ir s t bubble point" te s t. In th is te st, the filte r is w etted, follow ed by m e a su re m e n t of the p re s s u re re q u ire d to m ake the f ir s t bubble a p p e a r. T h is te s t had its f i r s t in d u s tria l u se to c o n tro l quality of po ro u s m a te ria ls by P a ll C o rp o ra tio n in the e a rly 1950's. F o r m any types of filte rs c o a rs e r than the b a c te rio lo g ic a l ran g e it p ro v id es a v e ry p re c is e in d icatio n of m ax im u m p a rtic le p a ss e d (1, 2). T h is sam e m ethod is recom m ended by m an u factu rers of p la stic m em b ran es fo r u se in th e b a c te rio lo g ic a l ran g e (3). Two p ro b lem s w ere experienced w hen we attem p ted to apply th is te s t to P a ll C o rp o ra tio n 's new 0 .2 jam ra te d AR e lem en t. F ir s t, w hen in d iv id u al e le m ents w ere w etted w ith w ater and subjected to v isu ally observed bubble point testin g , and the re su lts c o rre la te d w ith the re su lts of the sim u lated serv ic e te st, it w as found th at a b ro ad bubble point ran g e ex isted , in w hich som e elem en ts would p a ss and o th ers w ould fa il the sim u lated s e rv ic e te st; above this range all elem ents p assed , and below it all elem ents failed. Thus, a su b stan tial p ro p o rtio n of the elem ents re a lly suitable fo r se rv ic e w ere being rejected, increasing our costs. Second, and m ore im portant, when a visually bubble fre e elem en t w as w etted and placed in a housing, and then p re s su riz e d to the sam e te s t p re s s u re , a sig n ifican t flow w as m e asu re d a t the o u tlet of the housing. O ur fu rth e r in v estig atio n s included the follow ing observ atio n s: A. P all C orporation b a cte ria l filte rs a re m ade of fibrous m em brane. A w idely used co m p etitiv e filte r m edium , GS22, is of all p la stic m e m b ran e co n stru ctio n . Using this p lastic m em brane we w ere unable u n d er any c irc u m sta n c e s to obtain z e ro flow in s im ila r te s ts , even though applied p re s s u re s w ere le ss than one fo u rth those recom m ended by the m an u factu rers for testin g . F o r exam ple, w ith an assem bly containing six sq u are feet of GS22 m em b ran e, w hich is recom m ended b y i t s m a n u f a c tu r e r to b e t e s t e d a t 40 p s i, a p p lic a tio n of 10 p s i r e sulted in a continuous flow of over 5 m l p e r m inute of a ir. B. The sam e GS22 m em brane was w ater w etted and te sted as a flat disc in a n open jig . W h ile th e n o r m a l f i r s t b u b b le p o in t w as 55 p s i (3), o b se rv a tio n a t 10X w ith th e aid of a s te re o m ic ro sc o p e show ed a la rg e num b er of b ubbles, clinging to the m em b ran e su rfa c e , a fte r a 10 to 20 m in u te e x p o s u r e to 10 p s i of a i r p r e s s u r e . T h is o b s e r v a tio n w as quantified by m easu rin g volum e of a ir flow a g ain st tim e, w ith the re s u lts shown in F ig u re 7; the cau se of the o b serv ed in sta b ility p ro b ab ly lie s in the fa ilu re of w a te r to w et thp = <<.,T > * 1*1 -- C. W hen th e sam e GS22 m em b ran e w as w etted w ith V a rso l (a p e tro le u m fraction w ith su rface tension 1/3 that of w ater), observation w ith a m ic ro sc o p e a t 10X m ag n ificatio n show ed v e ry activ e bubbling a t 3 p si; in the V arso l these bubbles w ere continuously re le ase d fro m the filte r su rface. B ased on the ra tio s of the su rface tensions, this observation in d icates a 9 psi o r low er fir s t bubble point in w ater (see in fra equation (2) fo r re la tio n s h ip b e tw e e n b u b b le p o in t, s u r f a c e te n sio n , an d a n g le of contact. ) The behavior of the GS22 m em b ran e d e sc rib e d in p a ra g rap h s B and C w as con siste n t in rep eated te sts w ith varying lot n u m b ers, using specim ens obtained d ire c tly fro m the m an u factu rer, as w ell as o th ers supplied to us as a co u rtesy by a pharm aceutical m anufacturing custom er, who had in turn purchased them fro m the m em brane m an u factu rer. It was c le a r fro m the above that v isu al observations w ere difficult or m isleading, due to the sm a ll size of the b u b b les, o r to the long tim e fa c to rs involved. It w as re a so n e d that d ire c t read in g of flow s at each p re s s u re could be m uch m o re se n sitiv e th an v isu a l o b serv atio n , and in ad d itio n w ould be q u a n tita tiv e , e a sy to p erfo rm , and probably fa ste r. A pparatus w as then b u ilt to p e rm it quantitative m e asu re m e n t of flow vs p re s s u re drop fo r v ario u s fla t sh eet m a te ria ls . The flow detection device w as a c a r e fully calib rated G ilm ont m ic ro -flo w m eter, w hich w as used w ith a 1/20 sq .ft. 2 a re a flat specim en te st jig providing a range fro m 8 m l/m in /M to 2000 m l/ 2 m in /M . D a ta o b ta in e d i s show n in F ig u r e 1, f o r G S22 an d P a l l C o r p o r a tio n 's AR grade, and fo r the n u clear type of m em brane. The AR c h a ra c te ristic s a re plotted fo r single and double la y e rs; the double la y e rs a re m o re sig n ifi can t b ecau se a double la y e r is u sed routinely in both disc and c o rru g ated elem ent applications. Run w ith V arso l, these te sts w ere fa st, and the r e sulting quantitative data co n sisten tly rep eatab le fo r a given specim en. The follow ing is an an aly sis of the th e o re tic a l significance of the cu rv es of F ig u r e 1. C o n sid er a filte r c o n sistin g of a single c a p illa ry , of c irc u la r c ro ss section, w ith d ia m e te r = D (F igure 2A. ) The c ap illary is in itia lly filled w ith fluid; the p re s s u re P re q u ire d to eject th is flu id h a s b een show n to b e (4): P = co s (1) w here is the su rface tension of the fluid, and is the angle of contact betw een the liquid and the m a te ria l of w hich the c a p illa r y is c o m p o se d . If is z e r o , co s = 1, and the p r e s s u r e re q u ire d to e je c t liquids fro m a filled c a p illa ry is p ro p o rtio n al to th e s u r f a c e te n s io n . C o m p a rin g V a r s o l (f = 25. 5, = 0) to w ater (f = 74, = finite) we see th at the ratio -- 2 . < 7 4 . (2) P V arsol 25. 5 W hen the liquid w ets the c a p illa ry s u rfa c e , the angle of co n tact G enerally, the hole through a filte r m edium w ill not be a sm ooth c irc u la r cap illary . F ig u res 4 and 5 show scanning electro n m ic ro graphs of the AR fibrous m em brane and of the GS22 p la stic m em b ran e m edia, w hich illu s tra te this convincingly. We m ust, th e re fo re , d ire c t ou r atten tio n to openings of ir r e g u la r c ro s s se c tio n s, as illu stra te d in F ig u re 2B. A nalysis of random shaped p o re s becom es m athem atically difficult or im possible; how ever, it has been shown (2) th a t fo r a s p e c ific sh a p e of p o re w e tte d w ith a flu id of z e ro c o n ta c t angle, a relatio n sh ip e x ists betw een the d ia m e te r of the m axim um sp h erical p a rtic le w hich w ill p ass through it, and the p re s su re r e q u ired to e je c t the fluid fro m the c a p illa ry _ C onstant D = ---- p ------ . (4) R eal filte rs have num erous openings, not a ll of equal size, and are m ore nearly represented by F igure 2C. As p re ssu re is increased, flow in c re a s e s as each su cc e ssiv e ly la rg e r c a p illa ry opens up, and w e s h o u ld e x p e c t to g e n e r a te a c u rv e lik e th a t of F ig u r e 1, in w h ic h the ex isten ce of flow at a given p re s s u re drop c o rre sp o n d s to the existence of holes of a given size. If we w ere to apply our p re s s u re drop flow m e a su re m e n t te s t to the configurations of F ig u re 2C and F ig u re 2D, we should find th at fig u ratio n 2D has m o re flow a t a given p r e s s u r e d ro p than F ig u re 2C; this reflects the la rg e r num ber of la rg e holes in 2D. Had we m e a su re d f ir s t bubble point only, we should have been m isle d into assu m in g 2C and 2D to be equal, w hen in fa c t 2C is s u p e rio r, having few er la rg e ho les. Some im portant conclusions m ay be draw n: A. M easu rem en t of flow vs p re s s u re drop yields m o re significant inform ation than m easu rem en t of firs t bubble point only. B. R e f e r r in g to th e c u r v e s of F ig u r e 1, th e o rd in a te is a m easu re of the num ber of holes and the a b sc issa is in in v e rse re la tio n sh ip to the d ia m e te r of the h o les. P re c is e m ethods fo r com puting the num ber of h o les, and the a ctu a l p o re d ia m e te rs, a p p ea r n o t to be a v ailab le. It is, how ever, p o ssib le to m ake som e approxim ate com putations, and th is has been done as follow s: 1. Known o r e s tim a te d b a c te r ia re m o v a l c h a r a c te r is tic s w ere u sed to co n v ert the p re s s u re sc a le to an "effectiv e rem oval diam eter" scale. 2. C o n v e rsio n of th e o rd in a te s c a le of F ig u r e 1, fro m flo w /u n it a re a to n u m b er of holes p e r unit a re a can be accom plished as follow s: ( i) A ssum e that the filte r m edium co n sists of a s e rie s of p a ra lle l c irc u la r c a p illa rie s. (ii) U sing P o is e u ille 's law , and the effectiv e r e m oval d ia m e te rs deriv ed in the preced in g p a ra graph, calcu late the num ber of c a p illa rie s c o r responding to the m e asu re d flow data. Step 1 involves assu m p tio n s w hich, if in c o rre c t, could cau se the p o re d ia m e te r sca le to shift, but the re la tiv e p o sitio n of the v ario u s cu rv es would rem ain unchanged. Step 2 a p p ears ju stifie d by Jaco b s (5), and is p ro b ab ly c o r r e c t w ithin a f a c to r of four. The resu ltin g data is plotted fo r the la rg e s t ho les, w hich account f o r 0. 1% of th e to ta l flo w , in F ig u r e 3. W e s e e th a t p l a s t i c m e m b ra n e GS 22 h a s a p p ro x im a te ly 4 x 1 0 4h o le s /c m 2 in th e . 21 to . 2 2 /u m ra n g e , n e a r ly z e r o h o le s in th e . 14 to . 21 ^ m ra n g e , and m o st of its holes in a n a rro w band below . lS^um. F ib ro u s m em brane AR shows a som ew hat sim ila r d istrib u tio n ran g e, w ith a sp re ad ap p ro x im ately 0. 0 2 /um n a rro w e r than GS22, and w ith a b o u t 90% fe w e r o p e n in g s on its h ig h ra n g e of , 21 /um . D ouble lay er of fibrous m em brane AR has a still n arro w er pore size d is tribution ran g e, and s till few er holes n e ar its top lim it of 0. 2 m ic ro m eters. A t th is point, we tu rn our atten tio n to s ta tis tic a l c o n sid eratio n s w hich in d icate th at in b a cte rio lo g ica l filte rs , f ir s t bubble point is not a good c rite rio n fo r gauging absolute rem o v al. The to tal n um ber of holes in both m ed ia is com puted to b e a t l e a s t 6 x 10 / c m (if 0. 1% of th e flo w is c a r r i e d b y 6 x 10 h o le s , n th e n 100% of th e flo w is c a r r i e d b y n o l e s s th a n 1000 x 6 x 10 h o le s ) , w hile the n u m b er of 0. 22 ,um d ia m e te r b a c te ria c o lle c te d to clogging is (by a c tu a l m e a s u r e m e n t) a b o u t 5 x 10 7 / c m 2 . C o n s e q u e n tly , th e p r o b a b ility 8 of a slightly o v e rsize hole being challenged by a b a cte riu m is only about 1 in 1000. We saw p rev io u sly th at the tru e f ir s t bubble point in 0. 2 m ic ro m e te r elem ents is v ery d ifficu lt to d etect. We see now, th a t the tru e f ir s t bubble point is in addition not a good m easu re of elem ent quality, since elem ents w ith a sig n i ficant num ber of o v ersize holes can still provide a high degree of assu ran ce of long te rm absolute b a c te ria rem o v al. F ro m th is we m ay conclude th at m easvirem ent of flow at a p re s s u re co rresp o n d in g to a p o re d ia m e ter at the upper end of the p e rm issib le range w ill yield m o re significant data than seeking to m e asu re the la rg e s t single hole. It is p opular to c la ssify filte r s as "depth" as opposed to " su rfa c e " ty p es. F o r m any y e a rs, the p la stic m em b ran e types have been p re sen te d in the co m m e r c i a l l i t e r a t u r e (6) a s s u r f a c e ty p e s , w h ile th e P a ll fib ro u s m e m b ra n e s h av e been called depth types. The in feren ce has been th at depth types have wide p o re d istrib u tio n , and su rface types n arro w . We see fro m F ig u re 3 that in fact the p lastic m em b ran e type has the w ider p o re size d istrib u tio n ; the AR g rad e is su b stan tially m o re u n ifo rm than the GS22 grade, w ith resp ectiv e re la tiv e sp an s of . 135 to . 21 c o m p a re d w ith . 125 to . 22 ,um. R e a lis tic a lly , d iscu ssio n as to w hich is "su rfa ce " and w hich is "depth" type is idle; if the in c id e n t p a rtic le is 0. 5 /am in d ia m e te r, both a re su rfa c e filte r s ; if it is . 05 /im d iam eter, both behave as depth types. A n e a re r approach to a tru e "su rfa c e " filte r is the n u clear type, shown in F ig u re 6, m ade by etching openings through polycarbonate re sin film ; this type of elem ent has a w ider range than e ith e r the GS22 or AR, extending by th e sa m e c r i t e r i a fro m u n d e r 0. 1 jum to w ell o v e r 0.22 jum, We s e e , th en , that the "su rfa ce filte r" has the w idest pore size range of the th re e , and can fu rth e r conclude that th ere is no relatio n sh ip betw een pore size d is trib u tio n on the one hand, and w hether a filte r is of the so called "su rfa ce " o r "depth" type on the other. We indicated e a rlie r that the units of the a b sc issa w ere d eriv ed fro m m ic ro biological considerations. B riefly, these w ere: L o w e r l i m i t of GS22 w as ta k e n to b e 0. 13 m i c r o m e t e r s , on the supposition th at each grade w ill p ass p a rtic le s 1/2 its ra te d a b s o lu te d ia m e te r . T h is c o u ld j u s t a s w e ll b e ta k e n a s . 16 o r . 18 m i c r o m e t e r s . U pper lim it w as tak en as 0. 22 m ic ro m e te rs b e c a u se th is g ra d e quantitatively rem oves b a c te ria of this d iam eter; how ever, the s ta tistic a l co n sid eratio n s noted above p e rm it us to stip u late a 2 s m a ll n u m b e r (e. g. 100 to 1000 or m o re p e r cm ) of la rg e h o les to e x ist in an effectiv e ab so lu te filte r . On th is b a s is , the la rg e s t p o re s in the n o n -n u c le ar m em b ran es could be as la rg e as 1.5 m ic ro m e te rs in d iam eter. The p o ssib le existen ce of such w ider ran g es does not change any of the co n clusions, and we m ay su m m arize this portion of the p ap er as follow s: A. It can be shown sta tistic a lly that f ir s t bubble point is not likely to be a good c rite rio n fo r evaluation of the rem o v al capability of b a cte ria l filte rs. B. The p la stic m e m b ra n e s now on the m a rk e t have tru e f ir s t bubble points le s s than o n e -fo u rth of the a d v e rtise d values; n e v e rth e le s s, these p lastic m em b ran es of the non n u clear type have been shown to function w ell fo r rem o v al of b a c te ria ; it a p p ea rs th at due to sta tistic a l n atu re of the p ro c e ss the sm all num ber of la rg e holes p re se n t cause zero or an undetectably sm all num ber of fa ilu re s. C. The fibrous m em branes have n arro w er pore size distribution, when co m p ared w ith n o n -fib ro u s p lastic m em b ran es and w ith the n u clear type. T he data of F ig u re s 1 and 3 w as obtained using V a rso l, a refin ed p e tro le u m fra c tio n having a su rfa c e ten sio n of 25. 5 d y n e s/c m . W ater as a w etting liquid w as found unsuitable fo r the follow ing re a so n s: (a) T he n o n -fib ro u s p la s tic m e m b ra n e s co n tain a w ettin g ag en t, w hich reduces the su rfa ce ten sio n to as low as about 45 d y n e s/c m ; the d e g re e of red u ctio n of s u rfa c e te n sio n is v a ria b le , depending on how m uch has been w ashed through, etc. (b) C o m p a ris o n s m a d e u sin g w a te r a r e n o t v a lid , b e c a u s e n e ith e r fibrous n o r p la stic m em b ran es have z ero angle of contact w ith re s p e c t to w a te r. F u rth e r, fo r both m ed ia flow of d isp lacem en t a ir is tim e dependent, and no stable flow reading can be obtained. F o r exam ple, w hen GS22 is w ater w etted and exposed to a ir d if f e r e n t i a l p r e s s u r e of 10 p s i, flo w r a t e g r a d u a lly i n c r e a s e s , r e a c h in g 2 v alu es in e x c e ss of 2000 m l/m in /M in one to two h o u rs, as show n in F ig u re 7. By c o n tra st flow of a V arso l w et m em b ran e is stab le w ith tim e. ll TESTING OF PRODUCTION FILTER ELEMENTS: F o r production filte r elem ent testin g at P a ll C orporation, the u se of w ater is n early m andatory; for exam ple, in field operation when filterin g aqueous flu id s, the aqueous m edium itse lf m u st be used. Since the fib ro u s m em b ran es contain no w etting agent, and sin ce our c o n ce rn now is w ith the one type of m a te ria l (A R -dual la y e r) only, we u sed w a te r to develop the re q u ire d quality control test. The th eo retical considerations review ed above indicated that the te st p re ssu re should be equal to o r slig h tly above the u p p e rm o st ran g e of rem o v al as r e c o rd e d in F ig u re 1, w h ich is 1. 7 p s i in V a rs o l. T aking in to a c c o u n t th e ra tio of th e s u rfa c e te n sio n of w a te r to Y a rso l, a te s t p r e s s u r e of 5. 25 p si (10. 6"Hg) w a s s e le c te d . A t th is p r e s s u r e flo w is q u ite s ta b le , r e m a in in g c o n s ta n t f o r 16 hours or m ore. A new type of te st, known as a "F o rw ard Flow Bubble P oint" te st, o r m ore b rie fly as a "F o rw ard Flow " te s t w as developed. This te s t co n sists of w etting the elem en t w ith w a te r, subjecting it to a d ifferen tia l a ir p re s s u re of 10. 6"Hg in the norm eil flow d irectio n , and m easu rin g the volum e of a ir flow . About 10 m in u te s is r e q u ir e d fo r s ta b le flo w to b e re a c h e d a f te r a i r p r e s s u r e is ap p lie d ; th e flo w th e n r e m a in s s ta b le fo r 16 h o u rs o r m o r e . The flo w m e te r u sed is capable of m e a su rin g a ir flow s in the ran g e fro m . 04 to 2 15 m l / m i n . A r e a of a s in g le c a r t r i d g e i s 5 s q . f t . (.4 6 M ) h e n c e flo w s a s low 2 as . 08 m l/m in /M could be m e a su re d . We found that at 5. 3 p sid none of the elem en ts show ed low er flow s than ap p roxim ately . 3 m l of a ir p e r m inute, or 2 .65 m l/m in /M . 12 When an elem ent is w etted by w ater, an aqueous m em brane is form ed w hich h as an a re a of 5 s q .ft. , and a th ic k n e ss of only .006". An an aly sis of the cause of th is m inim um flow ra te indicated th a t the o b serv ed volum e of flow could be cau sed by the diffusion of a ir through the u n im p aired m em b ra n e of w a te r. T his w as co n firm ed to be the c a se by te stin g the s a m e e le m e n t a t 75 F u sin g a i r an d CC> 2 a lte r n a t e ly . R a te of flo w of C C ^ w as 48 tim e s th a t w ith a ir . The diffusion c o effic ien t of any n o n -re a c tiv e gas thro u g h m em b ran e is p ro p o rtio n a l to its so lu b ility , and the so lu b ility of CO2 in w a te r a t 75 F is 48 tim e s th a t of a ir ; th is c o in cid en c e d e m o n s tra te s that flow of som e 0. 3 m l/m in u te through the 5 s q .ft, elem en t is due to diffusion of a ir through the m em b ran e of w ater. C o rre la tio n betw een the fo rw ard flow te s t and the sim u lated b a c te ria r e m oval se rv ic e te s t w as accom plished in the follow ing m an n er. E lem ents w ere se le c te d to co v er a w ide ran g e of fo rw a rd flow s; th e se w ere te ste d by the fo rw ard flow te s t and then su b jected to the p re v io u sly d e sc rib e d sim u lated se rv ic e te st. The re s u lts a re shown in T able I. Note th a t no e le m e n t w ith a flow of u n d er 199 show ed b a c te ria p a ssa g e in le s s than 2 to 3 w eeks of continuous o p eratio n . In o rd e r to p ro v id e a fa c to r of safety , flow of 100 c c /h o u r h a s b e en e s ta b lis h e d as th e m a x im u m flow fo r an a c cep tab le p ro d u ctio n elem en t. F o r s h o rte r te rm te s ts , up to 96 h o u rs , a co n sid erab ly h ig h er stan d ard could be co n sid ered . S tatistically , the c o rre la tio n betw een grow through re su lts and forw ard flow b u bble p o in t d a ta is 97%, w hich r e p r e s e n ts a s u b s ta n tia l im p ro v e m e n t o v e r th e 77% o b ta in e d u sin g th e f i r s t b u b b le p o in t m e th o d . FIELD TESTING PROCEDURES: The "fo rw ard flow " te s t sy ste m adapts its e lf v ery w ell to in -p la n t testin g by the p u rc h a se r. Such testin g is done in e ith e r of the follow ing w ays: A. W here the elem en t is rem o v ed fro m se rv ic e , or is to be te sted p rio r to se rv ic e , the ap p aratu s shown in F ig u re 8 is used. The e le m e n t is w a te r w e tte d , 10. 6" of Hg v a c u u m a p p lie d f o r 10 o r m o re m in u tes, and the resu ltin g flow read . A pparatus fo r ev acu ating ten or tw enty c a rtrid g e s sim ultaneously is available. B. A second co n fig u ratio n , shown in F ig u re 9, is a conveniently a r ranged accu rately p re ssu re balanced pneum atic trough, used for "in-situ" testing of p re s te riliz e d filte r a sse m b lie s. A ir p re s su re (5. 25 p si) is a p p lie d fo r a 10' s ta b iliz a tio n p e rio d to the in le t of the p rew etted assem b ly , w hile the effluent a ir is d ire c ted by a 3 -w ay valve to bubble th rough the liquid in the tro u g h to a tm o s p h e re . The 3-w ay valve is then tu rn ed to d ire c t flow to the c a li b ra te d b u re tte fo r 30" or 60", and volum e of flow read . Back contam ination can be p rev en ted by using alcohol in the trough. OTHER QUALITY CONTROL PROCEDURES: The AR filte r m edium is very w ell bonded, the volum e of re sin binder being 60% th a t of the fib e r to be bonded. R e p e a te d shock h ig h p r e s s u r e d iffe re n tia l flow te s ts r e s u lt in no re le a s e of solids fro m th is fib ro u s m e m b ra n e . R elease of p a rtic le s d o w n stream of the elem ent, due to "bu ilt in d irt" is k e p t to e x tre m e ly low le v e ls by o b serv in g c lea n m a n u fa ctu rin g p r o ced u res, follow ed by a v ery high velocity w ater flush, w hich sim ultaneously c*^r~r\r\ n f-V, r * n , f ,, ^ ^ 4. ^ _ -T-T^ ~ 1 ________ _ . te s t of its in te g rity at high d ifferen tial p re s s u re . T his is follow ed by the "fo rw ard flow " bubble point te st, drying and packaging. L evels of dow nstream p a rtic le re le a se com pare m o re than favorably w ith com parable co m p etitiv e p la stic m em b ra n e types (8). O rganic co n tam in an ts in aqueous effluent a re n o rm a lly zero by the p o ta s siu m p e rm a n g an a te te s t. P r e s s u r e dro p and c le a n lin e ss a re checked on an AQL b a sis. E lem ent can be rep eated ly steam ste riliz e d w ith no lo ss of c h a ra c te ristic s. 15 TABLE I C o rre la tio n of P seudom onas G row Through T ests w ith F o rw ard Flow Bubble P oint T ests E lem ent S erial N um ber F6 H I 05 J3 170 PTM 15 J6 2389N H72 PTM 14 J79 PTM 2 PTM11 H97 PTM 24 168 PTM 20 98-176 153 J194 PTM 21 PTM 23 97-1 J4 B49 90-23 J23 J79 J75 167 97-342 J-84 98-65 PTM 4 J50 H67 PTM 8 98-77 97-201 B37 Forw ard a ir flow @ 10"H g, w ater w et, c c /h o u r 16 24 27 28 39 39 62 81 111 112 120 123 125 153 190 198 199 210 213 213 216 225 230 238 238 258 268 270 282 291 315 361 363 400 450 471 545 600 645 H o u rs on s t r e a m a t 1 L / m in . and 2 to 4 x 10 P s e u d o m o n a s d im in u ta p e r m in u te ___________ R em oved R em oved p rio r to p r io r to C lo g g ed , clogging clogging, w ith zero w ith z e ro due to grow grow grow through at through through tim e noted - 288 384 480 384 336 576 384 480 336 384 336 456 432 456 336 360 192 456 360 456 338 96 480 96 312 456 648 240 264 144 528 336 432 96 504 96 120 144 1 Q9 TABLE I (continued) C o rrelatio n of P seudom onas G row T hrough T ests w ith F o rw ard Flow Bubble P o in t T e sts Elem ent S e ria l N um ber J52 J68 117 97-222 F orw ard a ir flow @ 10"Hg, w ater w et, c c /h o u r 645 650 685 900 H o u rs on s t r e a m a t 1 L / m in . an d 2 to 4 x 10 P s e u d o m o n a s d im in u ta p e r m in u te ___________ R em oved R em oved p r io r to p rio r to C logged, clogging clogging, w ith zero w ith zero due to grow grow grow through at through through tim e noted 432 384 480 1 AIR FLOW, M L/M IIM / m , 1 * O, ,,,> ,hmig,, ^ ^ nudar ( w ^^ 18 Fig 2. Flow of aw through wetted capillaries. CUMULATIVE NO OF HOLES PER CM* OF FILTER AREA LARGER THAN PORE DIA. o04 o on <T> O OD PORE D IA ,BY BACTERIOLOGICAL ASSAY, Ml CROM FTFR * Fig. 4 Scanning electron micrograph of 0.2p. fibrous membrane; 10000X. Fia 5 Scan, mig lectron nucrograph of GS22-0 22 fi n o n -fibrous plastic membrane, 10000X. * 072573-7A 5.000X. 072573-7B Fig. 6 Scanning electron micrograph of 0 2p. nuclear type membrane ; IOOOOX. 10.000X. *t FLOW OF AIR THROUGH WATER WETTED PLASTIC MEMBRANE, M L./M IN ./M 2 TIME AT PRESSURE, MINUTES Fig. 7 Flow of air through water wetted GS22 membrane at 10 and 40 psid. Fig. 8 Apparatus for performing the forward flow bubble point test Fig. 9 Pneumatic trough apparatus for forward flow testing o f in-situ sterilized assemblies. R eferences: 1. D. B. P a ll: U. S. P a t . 3 ,0 0 7 , 334, 1 1 /7 /6 1 2. D. B. P a ll: WADC TR 256, M ay 1956 3. M illip o re A p p lic a tio n M an u al A M 202, p . 35 4. M. K antor, WIED. ANN. 1892, 4, 399 5. S. Jaco b s, F iltra tio n & S ep aratio n , Sept. -O ct. 1972, 525 6. M illipore B ulletin MB404, p. 7 7. Handbook of C h em istry & P h y sics, 42nd E d ., p. 1706-1707 8. E. K irn b au er, Q uality A ssu ra n c e P ra c tic e and S tandards fo r U ltipor B acteria R em oval F ilte rs, P a ll B ulletin A B800-1-72