Some Fundamental Properties of Asphalts Citation Gullekson, Ellsworth Eugene (1939) Some Fundamental Properties of Asphalts. Master's thesis, California Institute of Technology. doi:10.7907/mkd6-4x08. https://resolver.caltech.edu/CaltechTHESIS:12112024-234645502 Abstract No abstract. Item Type: Thesis (Master's thesis) Subject Keywords: (Chemical Engineering) Degree Grantor: California Institute of Technology Division: Chemistry and Chemical Engineering Major Option: Chemical Engineering Thesis Availability: Public (worldwide access) Research Advisor(s): Beckman, Arnold O. Thesis Committee: Unknown, Unknown Defense Date: 1 January 1939 Record Number: CaltechTHESIS:12112024-234645502 Persistent URL: https://resolver.caltech.edu/CaltechTHESIS:12112024-234645502 DOI: 10.7907/mkd6-4x08 Default Usage Policy: No commercial reproduction, distribution, display or performance rights in this work are provided. ID Code: 16917 Collection: CaltechTHESIS Deposited By: Benjamin Perez Deposited On: 12 Dec 2024 00:11 Last Modified: 12 Dec 2024 00:12 Thesis Files PDF - Final Version See Usage Policy. 20MB Repository Staff Only: item control page

SOME FUNDA.tl'.fEN TP..L PROPE:RTIES OF ASPHALTS '.Ihesis by Ellsworth E. Gu1lekson In Partial Fulfilment of the Requirements for the Degree of Master of Science California Institute of Technology Pasadena, California 1939 Table of Contents General Considerations • • • • • • • • • , • • • • • • • • • • • l List of Materials Investigated • • • • • • , • • • • • • • • • • .3 Transport of Water Through. Bit,unenQ •• . .. Wetting Angles and Water 'I'ransport , • • • , t • • • ·• • ·• , 6 Method. of Measuring • • • • • , •• Application of Conte.et Angle Data• , • , • • , • • • • 14 • 16 Diffusion of Water • • • • • • • • • • • " • •· , • Determination by Weight of Diffusion Cell • • ., • •• • • 17 .. Detennination by Use of Quartz Spring Ba.lane$• "' ConstTUetion of Quartz Springs• • • • , • • " , • • • 19 Prepare.tion of Thin Sheets ot Asphalt • • • • , • , • • 21 E.v alue.tion of Change of Diffusion Rate with Partial Pressure Gradient • • • 26 Results obtained with Quartz Spl".ing. Bal8.l1~0 , , , • • • 28 Correlation of Diffusion to Viscosity.•••••.• • 37 Discussion of Diffusion Mechanisms ii • • • •. • • • • • 4-0 Applications of Difi'udon De.ta • • • • • • • • • ·• • • 42 Surfaoe and Inte rfacia.1 Tensions •· • • • • • • • • • • • • • 46 ~ ♦ The Need for Knowledge of Surface F..nergies, • • • • • 46 .. . .. A Method Found to be Invalid • • • 47 An Indirect Method for Interfacial Tension. Chromatographic Analysis • • • • • •• ,. • • • ... • 11 , • ·• • ,, 49 Quantitative Fractionation of .Asphalta by Adsorption • , •• 52 1.wleeular Weights of Adsorbed Fractions • • • • • • • • • • 52 Asphalts and similar bitumens are ext ensively used as waterproofing .materials and as proteotive coatings on metal structures,. such as pipe lines., Praetieal experience has thus far governed the • selection o!' the type of bitumen to be used for a specifie purpose J but preference a so diotated have been subject to ehe.:nge, and are not quite consistent within looale or industry. Undoubtedly the le.ok of fundamental knowledge regarding bi'tuminous substances eon• tributes to the diversity of practice. To make available more in• formation or certain fundamental properties of certain of these materials was the purpose •O f this researoh, which was eonduoted a.s a project ef the Asphalt Institute. Investigation of basio physical and chemical che.re.cteristios of bitumens has received but scant attention compared. to the total amount of experimental ·work performed with them. Most; investi~tions have been concerned with matters of immediate practical importance, having to do with production problems, the establishment of' empirical standards and tests, and technical difficulties related to specific ~pplica.tions. Asphalts have been avoided by scientists as a subject ot academic research because they are materials of inde:f'ini te composition, with properties depending upon the crude base stock and upon their subsequent treatment. In view of the great economic importanee of the industry, a eorrele.tion of basic characteristics to behavior -under empirieal tests will be of benefit. Di:f.'f1cult1es encountered with asphalts and pitches in service present a picture of some complexity. Saturated roofings blister, metalij covered with a.sphal ts eorrode, and pavements disintegre:to.,._ Si.nee the greater number or the manifold applications of a.sph.al tic produot:s have to do more or less d.ireetly 'With the exclu:sion of moisture, knowledge of the essential behavior of asphalts in thft presence of ·w ater and other eomponents of the applieat5.on $hould. be 0£ value, It this behavior ce.n be related to other properties of the material, a direct basis will be obtained for the design of a product for a spee.ifie use. Materia.Ls Used Semples of asphalts and coal tar pitches were obtained through ·l:;he courtesy of the lJnion Oil 0Qmp8lly at Wilmington, California., and of th$ Shell Dewlopment Company at Emeryville, California. Code nunibaTs assigned to the various materials are given below; Pen. at 77°., Softenillg .. 100 g,. 5 seo. Point, °FT• Description l San Joaquin valley st eWll,,!o-blown asphalt 2 II tt 11 3 ff tt 1t ti 5 174 fl fl 2 165 ff ti 0 204 15 15.9 D-31.00.4 Steam-blown asphalt from crude source A 0.3100-5 fl u n II n II It 45 Ul6 D•3J.00-i tt ti t1 It ti ff u 7 189 E-5198-l ti ti It II n 11 ti l 234 E-519S.2 fl 11 !I II It ti ti 26 l40 E.5198..3 ti n fl fl tt ti ft 30 135 1 11')0 M-.l Steam-distilled t'rom Panueo orude, Moxie<:> .M•2 tt " It " \I ti 12 176 M•3 It II ft tt 11 tt 20 158 M-4 It ti 11 11 If {I 34 142 M•5 n It 11 1l II 11 46 137 24 166 19 188 Air-blown a.sphal t from crude source A 20 169 11 20 190 ll C:aUfo:rnia ai.r -blown asphalt 12 D.;.3100-1 fl " If II n ft ti 11 fl Pen. at 77° Softening 100 g, 5 ~c .. Point. 0 , . .D...~l.00.3 Air...-blown asphalt from crude source A 18 216 E-5198-4 n ti :ff 1l 11 rt n 20 235 :ru..5).98.;;;.5 l'1 " fl 11 tl . ~' 22 162 19 23..() 3 244 Coal tar pitch 24 14Q...l.50 2 l <jO,.J. 70 2 18~190 il Filled ai:r-blo-vm Calit'ornia asphalt 22 31 fl " " 32 II 11 fl 33 ti n ·n It fl 41 Coal tar enamel 2 160-195 61 New Type coal -tar piteh 2 21~ • Penett-aittQn. • The distance e. standard needle vertically penetratea a $anlplo of the material when the needle is loaded with 100 grwne fol' 5 s.e<,ondi e.t 71q1. • •• So~~I ~~.!:!;t, "" 1.he wmperature at whieh a steel ball will sink thir0ugh a pl'Uls of asph.alt held in a brass ring of suitable diameter. :Ct is :i:·easo:ned,. fi'Om a. knowledge ef the eondi:t"us under which asphalts fa.il to exclude llre.te r~ t hat transport through the p~tective layer may be aeeo1uplished in one or more of three ways, 1. Fe1i.etration t hrough microscopic eracks or pin-holes by the action oi' surfaee forces; 2. B.lectl"osmosis. a.e a. reeu.lt of electrical po'ventia.l dif..,. :t'erenoes across the ooatingJ and 3. Diffueion und.e.r the iµ.fluenoe of aqueous tension gradient1!!. f:'4:r-tial in~stigation$ of each of these three have been conducted. A liquid in contaot with a solid assumes a surface form which permits a statie balance of the surface forces existing in the inter..faoes which join_. .From Figure 1 it is seen th.at a force be.lance at point A gives the equation Liquid /{e Il~~~ ~ Figure I r;__ -r;L A knowledge of the magnitude of i, the contact angle., is sufficient £or a qualitative :measure o!' the attraction of the eu;r;faee ~tf the solid fox- the liquid. I:f' Q is o., tho liquid tends to spread indefinitely in a thin film over the surface {Jf the eolid; and if ~ is l80~, the liquid will not wet the solid at all. Sinee the single quautity Q is in the case of solids :more easily measured than the surface and inter.f'aoial tensions., it is used as an index to the tendency tor water to mow into ere.ck~ or pin...holes. Method and 0.l,)fa..ra~sl • Several methods of measuring cont~ aot angles have been developed and used. Talmud and Liebman (l) used with success e. microscopic method for de t ermining the dimensions of drop:;1 of liquid on a flat ple:ce, and calculating there.f rom the contact angle,. Ablett (e) constructed a cylinder., coated with the solid to be studied., which could be rotai;ed in a ·!;rough containing the liquid. Hi.a procedure was to rotate the cylinder at a eonstant speed., and raise the level of the liquid until the surface was horizontal up to the point of' contact with the cylinder. Measure:m.ent of the liquid lewl in relation to the cylinder a.i'.ford.od data for calculation of the contact an&le,. Adam aud Jessop(S), a.nd subsequently Nietz( 4 ) used a method wh~reby a ooa.ted plat e was tilted -while in the liquid until the meniscus became horizontal to the point of eont aet. Since this method was reported to be as reliable as e.?J¥,. it was selected in preference to the others 'b ecause of the ease with which it could be applied to asphe.l ts. A tilting plate appar.e .tus, shown in Figure 2; was construeted. It eonsists of a rigid stand which supports~ reote.ngule.r trough CONTACT ANGLE A , 0 PARATUS Sca/e - Ha l r 31.zc :1 1' 'i··1 n· " ! I I ..5crew r o r ra1s1n9 o/?d / ower// 7!7 s /1de Spring r-- -- , I I I I I Slide cool'ed w/th ospholr - ---~ 0rcu/or sco/e groduoled 1n degrees Trough ror solvrion Knob ror rol"ol'ing scale ond slide - -- Counlerwe/ghr Leveling screw FRONT VIEW Ftgore Z S/0£ VIEW .fitted with plane glass sides. By means 0£ a.n aoeurately machined. alide and a. screw, a microscope slide coated with asphalt :me.y be moved 1n or out of the liquid. An eye-piece is u~d to provide an enlarged view of the junction between the surface of the liquid and the asphalt coating. '.lhe assembly which holds the slide may be rota.ted about the optical a::d.S. of -t;he system, and the degreetl of rotation may be read on an ai'fb:ed graduated scale• Theoretiee.Uy, if' a plate il'J immersed in a liquid., the true contact angle should be obs~rved if' time enough 1$ provided for the attainment of equilibrium. However, the time required is so lon& that it is almost impossible to measure t.he angle under e.o nditions of true equilibrium., Investigators report the.t equilibrium anglea are not even approximated after periods of waiting as long as 24 houre or more {5 ) • It becomes necessary; then,. to measure contact ang les more or less dyruunically; and to correct £or the lag in approach to equilibrium. Anglea have been eomi.nonly observed to be larger when the solid is im,rn.ersed in the liquid than when it is being 'Withdrawn. Such a. phenomenon sugge,ats a "hysteresis" effect which Adam (o) a.,cribe, to .friction between the surfaces. Some controversy h~s arisen in regard to the ea.ulje of the lag., but the derivation of the equilibri'U!ll contact angle from immersion and reeeding angle& is quite generally aeoepted as explained. by Adam. Let F = the torce of nfrietion11 which just p:revents relative motion of the pho.$e s,. ~i = immers.ion eontact angle• ~r = reeeding contact angle T = s,Ul".face te nsion oi' the so lid 0 TL~ surtaoo teneion of the liquid T = interta.ciaJ. tension between solid and liquid SL Q :::: equili.b,1•imn oontac't; angle At equili'brlum there is no :f'oroe oi' .friction;; so (l.) 1'Vhen the plate is imrner3ed (2) And when the plate is withdravm. Addin,g (2) and (3); 2(T8 • 1'61 ) = TL (cos ~i + coa gr) Substituting in (1), eos Q = cos Qi+ co$ Qr 2 Aceo:rding to restrictions placed on th.is derivation originally,; it is valid only ~ case there i .s no motion of one phase rela:tive to the other at the instant ot measurement., (2) However# Ablett • has iahown that if receding and immersion angles are determined at the same rate of relative motion of phases, the relationship still holds. I With the apparatus u$ed in this reaeareh, the juncture of liquid a.nd sol:l,d wa.s oe.used to oceur at the a.xi$ of rotation of the slide assembly when the meniscus is horizontal. Careful :manipulation then :r·esults in tl eondition quite olose to static at the titne when the angle is read. However~ readings made while the slide was slowly meved in o:r out o:f' the liquid gave results which cheeked the others- w:t thi:n the ~xpected error ot the ste:tio determinations,. The slides were coated with asphalt by dipping into molten asphalt, allowing to drain, and then to cool in air. Exerimental Result~? ""' ~ test the roprodueibility of the reading~ on eonte.ct angle&, three independent sets of determinations 1vere ma.de on six of the asphalts. '.ihe maxiw.um discrepancy bet·ween flIJiJ- two of' the d,,ter-.1'.nin.ations for one type of asphalt was found to be 1. 75°, and the gen~ral agreement was much better than that, llie data indicated that if a reasonable number of independent measurements on the s.nme sample are made, the average ought to be consiatent with respect to measurements on o·cher samples within 1 °. At the outset there ~-as some question whether the angles measured were of the correct absolute value. Slides t'10re coated vd th pe.raf'f'in wax. and eontaet angles detennined for fresh distilled water on paraffin. 'Ihe average of the v-.alues obtained was 108°, 11'1'<1ioh agrees quitt;i well wit..1. the 105° reported by Adam and Jessop (5) ; 4 and the value of 109° j; 30 1 reported by Nieti/ ). Care vra.s ·taken to maintain a clean fresh surface on the liquid by frequent cleaning of the vessel., flushing with £reijh solution before each reading,. and siweping the surface with e. paraffined .float. No discrepancies vrore noted \mich could be at tribut,ad to the presence of a eo11taminating surface film, an observation al so reported by Niet,(4:) ' . The dependence of the e.n; le of contact ::-n tho pH of the solut.i on was investigated. To elixainate variation in ionic strength bettw,~n solutions of different pH values, Clark and Luba 'bwl'i'ers were prepared ar.id brought to equal ionic strength with Na.Ol. fue surface tension$ of the resulting solutions were :mettsured v.rith a Du Nuoy tensiomete:ra and the pH values were determined with. a glass eleotrode pH meter, Data obtained are as follows; Ta.bl~ I pH 0.99 1.93 4,96 7.16 10,99 T, ~iynes/em.. 62.6 62.3 67.,9 57.,6 59.-4 1 85 79 '76 86 82 2/ 81 80 56 ( 'l) 85 79 3 84 74 75 77 80 11 97 95 96 94 95 12 97 96 96 95 96 21 101 96 91 99 100 22 87 86 95 93 90 31 50 44 53 40 21 32 5-8 ::.o 55 55 31 3b"' 64 57 62 52 64 41 67 66 66 60 61 .Mate rial Bach val ue recorded in the tabl~ is an average of at least f our read• ings on each of two or 'J.IOre slidQs coa tad with asphalt. '.there seems ·t o be no signii'i.e ant va riation in wetting angles over the pH range or these buffers., except tor the t wo ooal•t.ar In order to obtain scm1e data. cove:tin& the behavior of wetting e.nglee in a.:;ids, bases, and salts, soluti~llS oi.' HGl, KOH, e.nd CaClsa ware prepared. CnGl2 was selected oooauae it afforded e. 10..rge range of attr£a.ee tendon of solution, and it wa.$ thought desir~ble to clis• eover whether the surface tension ot the $Olution would have a marked eff-eet on the contact a~le., Table II. Data for these solutions s.re given in Table II Concentrntion 0,05 N 1.29 N o.5 N 1•.1s N o•.04 N 0.1 N 20•.0 N H~O HC1 HCl KOH KOH OaOl~ Ca.Cl2 Ce.012 n~.o 72,-6 72 4 • 73.8 l 89 83 90 66 2 82 88 86 68 42 80 3 78 82 86 63 40 80 11 97 98 $3 80 6$ 12 100 99 92 '77 74 21 100 $9 lOO 83 22 91 97 92 31 74 68 32 73 33 41 T 76.). 5o.o tsl.,.O lOo.O Sample 96 100 73 96 96 107 93 97 108 80 98 99 110 78 73 96 96 108 69 31 2-8 70 72 63 25 28 66 73 75 67 20 24 68 76 '16 66 21 24 Sl A marked decrease in the we tting e.ngle was obse1,red in all ea.see in KOH, although the ooneentrt:"l.tion required to produce the decrease is less for eoe.l to.rs than £or asphalt~. The slight i,n.... o:rease in a:rigles for t.he coal tar pi tch~s in l. 3 N KOH may not be re·a l because of the di:f'£ieul ty of measuring; small angles. Goal t a r pitches.; in contrast 'bo the a.sphaltsi showed a definite d.eorease in wett i ng angle in Ii Cl solutions. The contact e.ngles in Gs.Gl2 solutions appeared to have no marked relat ionship to the concentration. exeepti.ng that :for :w.o N CaCl2 the angles are larger tor asphalts; and in general slightly smaller for the coal tar pitohes. Apparently the solution~asphalt inte rfacia.l tension increases with oonoe:rrtra tion slightly, and the solution•eoe.1 tar pitch inter.facial ten sion de0t-eases $Omewhat.ihe changes a.I'e, however., not 8ignifiee.nt; since a n angle around. 90° would have to change appreciably to indicate a sub sta.ntial change in inte rfacie.f t ension. Disousdon 0£ Result$: "" Water and aqueous solutions., being polar, woul<;i be e~cted to have g re atest affinity for polar surfaces. On this basis ·!:'.he su.rfe.oe polarity of the classes of ma.:te rials may be .ran..°i<ed in the following order oi' dec r easing polarity, coal tar piteh > steam... blown _a .sphalta >air...blown asphalts. It i s quest ionable, hovrever. to make this a eo njeoture concerning the average eh-einica l constitutions, since there is evidence that,. for i n stance• the oily portion of air.. blown aspha.1.ts exudes, or 11 blooms 11 on the surfa.ee,. mnking the surface eharactoristios a .function of the dispersing phaE?e. To illustrate the applicat ion of contact anl6 le data. the level$ to vJhieh tv-ate r will rise in asphalt and coe.l t e.r pit.eh ooati:nr;s will be computed.. If we assume e. vart:!.eal oireularly•cy lindrical hole of radius r in a thick coating., the height to wh ioh a liquid of' density d and surface t ension 1' will rise is c; iven by the equation 2T oos =---dgr Q h As a specifie example take r = 0.1 mni. He is;;!:t t, h ., i~.a t';I ri,al . No• Co1J.t~et, 5le, ~ l 88 o.oo 2 $2 2.04 3 78 3.,06 11 97 ..1.ao l2 100 ... 2.. 5Ci 100 ... 2.66 96 ...2,04 31 74 4'.06 32 75 4.30 35 73 4,.30 41 76 3.56 Qm..• Water will not penetrate the air-blown aspha.l ts, but will penetrate the s te.e.m.... blown asphalts and coal -t;;ar pi tehes to a height of several cent i meters. Similar calculations could be made for eraoks. Thi.s investigation was carried out: on fresh or compa1·atively fresh surfaces. It is conceivable that weathered asphalts would show a. substantially lower contact atl/)_;le, since water.. soluble constituents a;r e form~d. However, the surfaces in a era.ck in a. coating would re,.. main relatively .fresh for some time, einee the principal e ffec·c of weathering is due to photochemical action. kn interesting supplement to -these .,:easm•e1w11.ts would be a. study of the ef.feet o.f weat,hering on tJw we tting angles. Eleotrosraosis experiments were conducted by Charles Best,; Asphalt Institute Resenrch ASsi$tant, 1936-37• '.fhe results indi"" eated that water transport through asphalts by means of eleotrioal potential dif'1'erenees is nil. If e. eonst1mt a.queous tension gradient is maintained across a sheet of' material, the weight,. w. at water difi'used through in time., t., under steady state conditions will be e_;iven by F'ick's law o:f d.i f'.t'usion, w= where KAtllP L A :;:: area of difi'usion pa.th. L = length of dif.fusion path. liP == partial pressure gradient over the length., L, of 'the path. K == dif.t'usion constant. '.fhis relation applies to a.,"ly steady state di.t'fu$ion proces s if ve.lues of LlP a.."ld 1 rems.in oonata.nt, and applies tmrestriotedly to a case in whioh the part;lal presfmre varies linearly a.long the diffusion path. If hygroseopio materials are . p,?'e$ent in the diaphragm. or ii' adsorption of ,v:ater occurs., the concentration of water as vapor in the membrane \vill not 'be linearly related to the vapor pressures in contact; and the linear di.ffu.sion law constants will be found to vary with thJekness of diaphregm and with w..por pressure gradient. A true deqei•iption of the diffusion process under such eomplicating conditions 'i.W.Y be obtained by 1;he use of a r.1.0.re 1.nvolved expression; but i'or pre.ctie$.l pn:rposes Fiek:'s 1~-w :ts a oon.wnient basis fo:r eel• eulation., slnoe the quantities :i.t involves a.re :rei?,dily measurable. For MY wd:;erial, ad0que.t-0 inf"onoo.tion for the compu't:ntion of dif..f".lsion rates nll.l.y be deriv·ed from esta.blishml.mt of the variation of linear dif.t.\,sion loov const11nts w-lth pres$1.1.re grad:tonts Ei~d with thicl::ness of' material,. :Methods and App~ra~s: .,. Measurement of the diffusion rate was accomplished by a simple and direct method. Constant aqueous tension gradients were maintained. a.cross a thin sheet of asphalt, and the amount of water transported through determined by weight as a function of ti.me. '.Ihe weight of water diffused ·through was obtained either by loss of' weight of we.tar f'rom the high vapor• pressure side, or by e;ain in weight of' a desiccant on the low vaporpressure side. The first determinations of diffusion rate were performed with a simple type of dif!'usion eell, consisting; of a spun anodized aluminum cup with an al·un<.lum :f'ilt,er dlsc support for the diaphragm 01· asphalt, which vms sea:led o\rer the mouth of the cup, as s..11.own in Figure 3,. Alundum disc Spun Al cup Figure J. Oi/rus1on Ce/I '.Ihe lewer surface ot the thin diaph.rag;m was $ea.led around the edges to the eup by ca.:reful application of heat, an.cl the difi'usio-n a:re-a WQ.S :restricted on the ~pper surface by centering on it a heated dural ring, '.I.he space between the lip or the cup and the du1·al ring ws..s poured full of ptu•af'!•in wa.:x: to help prevent leakage. Blank runs indicated th&t this was (irt ef'f'ective seal, Thro types of experiments were oonducted with the cups 11... lustre:ted • in Figure 3. In one case the base of the .receptacle was pa.rti.e .l).y i'illed with -vro.ter before s.a aling tht: diaphra,yn in place, Ut11 the assembly placed in a. desieeator in a constant tempera.... ture water bath. The oup.s ware removed periodically and weighed. T.he other method was aim.ply to reverse the position of' wr:d;or and d.r ying agent, placing ·w-a.ter in the dtH3ioeator, end the dryine; a.gent :i.n the cup. Diffu~Jon of water through asphaltio :i:aaterials proved to be relatively slow. The time required to establish an ac.cu:rate steady• state rate of diffu.s ion was i5teat because of ·the errora involved in obtaini:o,g ohru:iges in vreight a& a. SJaa.11 differen.ee 'between two com.paratively lt:Lrg;e figures wh ich were in themeelvef3 subject 'to accidental eha.!4-:!;e~ incident to .frequent handling. Resu.lta 1ndieated the desirability of' :finding a method whereby sri1aller s:m.ounts oi' moistu1·e could be determined wi.th mode:rate precision, Ji different type of oell.- similar to the one used by Taylor, 7 Heri'!..nann., and Kemp,( } was desi 6ned to avoid the objections found to the methods outlined above. In eesenee th.is new cell., shovm in l~igure 4-, di.f'f'e:rs from the othe:rs only in that a quartz spring 0/FFUSION CELL 5cole-Fu// Size ( Pyrex /'ube - __ ~ ~ ~ ~ -==... ~ Quartz spring - -- - - ►~ CoJ:)1//ory pressure equo/1zer . . , ,-<------ --- -----... --- -- ~ Reference point __ I -=----=====-~ ~ c:::::::=___ ~ Anhfjdrone ----- - -----...._ rbro111n seo~ -------- Wax sea I ----------- ◄-A lurr11num cu,,o Solution Figure 4 . halo.nee for mea.1,mri~ cha.1iges in weight; oi' the drying agent is added as a aupe,rstruetl.'l.re to the cup. Wa;\:;e r passes i'rom the 0up at the bottom., through the diaphragm ,. and into the spa.c~ above., vme re i t h absorbed by Mg(Cl0 4 ) 2 held on ·a n a.l'Ulllin.UL1 :t'oil pan supported ·by the e~librated fused (ri.1ar-tz spring. With this eell, iMall a..rnov.nts 0£ we.tar are EH,~sily deteetod wi:thout dtsturbing the sys-tern. A solution for maintaining conste.nt h:wnidi ty was plaeed in the cup, an asphalt diaphragm sea.le-d to an accurately turned and ground i'lat brafis ring w:as pl!'.i.cod over ·the mouth of the eupt and over t his the inverted test tu'be containing the qua.rtz sp~ing supporting some anhydroM was sea led with a t h ick la.yer of pe.re.ffin, '.lhe as sembled cells were placed in an air thermostat maint ained within 0,,2°C. of the desired temper,~ture,. 'Ihe elongfitions of the springs were 1:1easured with a cathetometer without remo;ring the cells from the t hennos·tat. rt was found in some ease s tha.t tho absorption of i-;he wate r vapor initially present i n the inverted test tube resul ted in a presaure differential across the diaphragm sufficient to ea.use it to war p o:r to i'raeture. To avoid this a ca.pillary tube was pulled out on, the side of the eell. After assembling the cells,. the aa p:i.lla.ries were left open for three or four hours, and then sealed. In all cases whe 1·e this proeedure was follow➔ d there was no druna.6 e to the ~sph'.ilt menbra:.1e flS e. result of pressure reduction. Oonstruc~ion of Quartz Sprinsass • In order to construct th.e se cells» it was necesisa.ry to make qun.r'ta; springs, since none were readily ava.tlai..)le or1. t he 11'.!s.rket. A method and appaartd,us :fo:r making quartg h~lioes with ea, se wa.s developed._ Aft{: r som,e expeTl.m.en:tn tion_. a machine for Winding spring$. trow pre..f'orm~ci fibers was constructed. 'lhe essent ia.l de t ails of the app:ua tus a r e ahowu in Fiiu:ce 5. li'ibtn•.s were prepared oy he-a.ting a section of a. clea.r qu.arti. rod, in an ozy""hydrogen i'lo.me, (md dre:wing 1:;he fused portion down to e. fiber e.baut 0.08 rmri.. in diamet er. With a cert,ain amount of p:ractic,e to d ~velop the technique., one is able t o prepare fibers of the prope~ 1 size ru1d ot rolatively unif orm diameter t hroughout their lengths. '1!o con st ruct the spring, a fiber is rwi through tht$ oe.p.: Ulary fe ed, a hook is formad iu the upper e nd with the a id of a torch, and aff'ixed to the wire support. On a dmilar hook ri.t t he lower end of -t.11.e fiber is placed a. weig ht oi' about O, 5-.l.O grams to assist in kee.pin{; tension on the s pring while i t i s for.med on the silioa mandrel., After the fiber is thus S<: cured., the motor, ad.- justed to t;~rn th.:1 (1andrel at a.bout 3-6 r.p.m., is started~ and the torch wit h a l;as .. oxygen flame is swung around and direoted on the fiber just above where i t. ·touches the ,;1a..YJ.d:rel. As the. fiber so).'ten s a.11.d confonn.s to t he shape of t he quartz tuba, t he ca pillary fee ,l e.nd ·t;ho torch are advanced b:y the scr~w· on which they are mounted. Any desired pitch of helix may 'be obtained by varying t he size o.f the pulley which drives the t'hre ad1;;,d she.ft. P1·oper adjusti;nen-t of th.e i'lrune and o.f the speed of the motor will result in perfe,ctly true helices., SL'loe the translucent qUe.rtz tl).l)ing used for a manditel has a. slight ta.per as made., there is lit-t;le OEVICE FOR WINO/Nu OUARTZ SPRINGS Quorrz he/ix t;Juor/z moadrel----'~ Threaded short" Torch Co,tvl/ory reed - - ~ t;Juor/z r1ber -~~ Fi9ure .5. d:tffioul ty with :removing the spr1.:ng:s a fter they t.i re f'o.rmed T.r.e springs ·were ca libra:ted by measuring the elongation w:i:th kno1.,.n wd.ghts on the foil pan. Over a wide range they i'ollov~rea. Hooke's h:w withj_n the e1·rors of meast,rement. Se.nsitiVities oi' the !;l pring s used varied between o. 5 and 2 •.6 milligrams per mHli• meter of dd'leetion. Pr~P_.~ro.ti?~ of .AsJ?halt Die.P!1:l'a,IQllS: • Some diffieul ty was at :first $xperieneed in preparing thin test sheets oi' asphalts and pitches. 'lhe e~treme britt;leness of some of the samples made necessary the development Qf methods and teehnique in manipulation; and depressions in the surfaces of the diaphr~qn invalidated m,ea.sure!Ut3nts of the thickness. Asphalt v.ras known to contain water and entrapped non-condensible gases. From observ·ation of its behavior on heating , and from the appearance of freshly f'raat-ured surf'a.ces, it was decided t h~tt voids might appear in the test dinphragn: unless s t eps were taken ·t;o remove air and water. Indeed, preliminary diffusion runs indicated that some such imperfection v;aa present. Removal of the air and water was aooomp1ished by heating a. small sample of the material to a.bout S0°c. above tho sof'ten:i.ng; point in a f'laak in an oil bath, and applying; suction with a water aspirator until bubbles ceased to form. '.lhis usually took about 30 to 45 minutes. '.!hat this process removed the occluded .:.sases wns shown by :results of 11 void t ests" before and e.fter the evacuation treatm~ntJ :Pigur.e 6 shows t he apparatus :for this test. P~ t-ube is f'illod wi:bh Stopcock Tube rilled with solve17/-. A ~phoIr -----1-t-++---, Figure 6. f/oid Te.5/ Apparatus. a solvent; such as benzene for asphalts and carbon disulfide for coal t ar pitches, and a second tube containing the sample on a screen is lowered into it. The solvent is allowed to rise in the inner tube until it reaches the stopcock, whiab. is then closed. As the material on the screen is dispersed in the solvent., entrapped gases are set free., and they collect in the capillary tube at the top, Tests with a number of different samples as received indicated the presence of an appreciable quantity of "voidsu; but evaeua ted se.r.1ples subjected t.o the same procedure invariably showed an amount scarcely detectable. It v1as thought that, because the properties oi' asphalts a.re subject to change on heating,. the evacuation treatment might alter the material substantially, In order to determine whether this was the case, it was thought advisable to select a property known to be variant with heating, and to :meas'llsre the effect. if any. the treatment on ·that property. ot So.fte:iling points, d-etermined by the standard "ball and ring" method, w~re thought to be sufficiently indicative. and ·were selected as test oharacteriatics because they are e a sily dot 0rmined, Softening points measured before and after evacuation were the same within experimental error • two or three degrees Fahrenheit ... and agreed within a like amount with softening points obtained by the same procedure in the laboratories of' the co:-:porationa which provided the sa:rnples 11 As an additional precaution 'before a great deal was done with the evacuated asphalt, diffuaion runs were made with asphalts which had been subjected to various degrees of heating and evacuation. Some samples vrore heated under vacuum for periods appreciably .) .onger than necessary to remove voids., and their diffusion cGnsti.;.nts determined, Since the values obtained were not significantly different from those determined for the $a.me kind of asphalt which had been heated and evaeuated for ~ minimum of tim~, it ,val!I concluded that the properties of' the ~sph~lts «ould change no mo~-e during heating and evacuation than the wnount by which they vary fro~ point to point in a batch of asphalt. 'lhe samples after evacuation ·were east, for eonvenience 1 into $lllOOth pellets on warm mercury, ea.ch pellet containing enough for one diaphragm., The method :Una.Uy selected i'or forming satisfactory test sheet$ was to press each of the pellets to a definite thickness between two flat amalgamated brass plates placed in an oven set a few degrees aoove the ooftening point of the asphalt. 'lh.e ·. apparatus with ,IJili ch t his was aceompl:t~ed (1''igure 7) consists primariiy ot an arrangement whereby t wo springs ma.y be used to force the two plates together while they are in the oven. By means 0£ the bolt thr-eaded through the uppermost cros~1uember., the springs may be compressed before the assembly is heated,. '.the t wo bTa.se plates, in t:ifder to insure that the two surfaces of the diaphra~ would be parallel, were a.ceurately turned flat, and then ground -together before runalga.mation. lhe thickness .o f the finisheQ; sheet was regulated by interposing pieces ef flat brass stock, o:f equ.al thiokness, between the plates at three points around the periphery~ llie apparatils ·was removed from the oven as soon as the plates were se pa.ra.t ed only by the thickness of' the spacer&~ A circular portion t o use in the diffusion cells wns cut from each diaphragm. with a oireula.r emalge...nated brass knife edt5e, the diameter of which was accurately known. '.Ihe temperature at v.hieh the eut ting was done ho.d to be controlled properly in Ol"der to obtain a disc without fracture. Since asphalts. and pi.t ches do not adhere to mercury, the discs were easily removed from the plate i,f the temperature was maintained above that 1 t which the zna.terie.J. "beerune brittle. 'llle diameter of the circular portion wa.$ very close to the interne.l dis.mete.r of the cutting edge, so the area or the disc was easily computed.. ',lhe thickness of each diaphragm used was measured, since they were found to be variable in thickness by a fevr percent. Arno/gomo.red .brass p/o/es 0 F/gl./re 7 fJiophrogm Pres_y_ Densities on all the evacuated saraples mre obtained by a loss of ·weight in water method; and from the density of the e.sphe.lt and "the v,reig;ht and area of the disc the average thickness was computed. 'Whenever practicable this calculated value was cheeked by tt;l.king the average of a number of mior11nneter readings taken around the edge of the diaphragm. Usually the two independent measureraents checked within Qne peJ'cent. Variation of partial water va;eor pressuroi • A number o:f saturated solutions in contact with a de.f inite phase of the solid solute are suitable for maintaining eonstant humidity. Some of these were utilized as a simple and direet means of obtaining different partial pressure differentials in measUJ"ing the efi'eet of pressure gradient on the linear law diffusion constants. Solutions ot CaS0 4 •b1I2 0, NH4 Cl, Ca(N0 3 ) 2 •4H2 0, and CaCl2 •6H2 0 were used to maintain aqueous tensions of 23,3 1 18,8 1 11.9 and 7.31 mm. Hg., respectively at 25°c.; and , for measurements a.t 35°C. solutions of NH~Cl, which have a vapor pressure of 34,2 mm. Hg.~ were utilized. Magnesium perchlorate was used in ea.oh case on the other side of the diaphre.gm. An alternative method of measuring the diffusion rate unq.er cons'Gantly varying vapor pressure gradients was originated., and apparatus for measuring it and equations for interpreting results developed. If the water vapor available for diffusion is restricted in amount by having as the source a constant volume space containing only water vapor, as shown in Figure 8, the partial aqueous pressur~ in the space may be related to the amount ot water diffused through the m.embrane as described below. Ouor/z spring bo/ooce --~ Youree o/' 1ni/io/ oqueous Cup con,oia/ng d/op/7rogm~---. . . . T e n 7 Vo,oor 5poce Figure 8. J/or10.ble Pres..:5ure Ce// Let K = linear law di:t'i"'usion constant L = length of diffusion path A i:: area of diffusion path P = aqueous tension., rma. Hg., t = time =volume of vapor space w =weight diffused through at time t V m =wt.or water vapor in vapor space n = conversion factor bEitween vapor density and pre~sure = 1.,03 :x ic/J for water at 26°0 If the vapor space is so large that changes in aqueous tension within it oeaur slowly., i-t may be assumed that the unsteady... st ate diffusion takes plaoe under oouditions olose to equilibriwn• '.lhis is thought to be true because test diaphragrJ.1$ p-rsviously exposed to normal atmospheric hu.'1lidity require only a. few hours (12 to 15) for the establishment of equilib1•ium, as judged by the approach to linear diffusion rates., when plaoed bet."reen a partial pressure gradi .z nt of 23. 3 mm. Hg. Then unde r oondi tions of' slow change of vapor pressure in the vapor space, f'or an increment o:f weight dw, dw = KAP L • dt At any instant in the vapor spaee; Il n>, P=V ltt any time the amount of v,ater absorbed by the drying agent on the spring will be the sum of that which ha.s diffused through and that which has left the diaphragm because of a change in concentration within it as demanded by the constantly varying pressure gradient, Experiments on the absorption of water vapor by asphalts indicate that ·the moisture whieh could be lost by a diaphragm of the size used would be quite small. So if the smount of vapor dif.fuse.d through is large compared to the loss of moisture from the diaphragm. we may £or practical purposes neglect the la'i!;ter. X't1E1n m = ffio - w~ where U\, is the weight of moisture initially present in the vapor space. Also F = P0 ..,. wn V \me:i:-e P0 is the initial aqueous tension in the vapor space. dw = KA P0 L · dt... K Awn LV dt !! ::::KAP0 ... KAwn . dt llie quantity L . . ' . ,. . . LV dw dt is the slope of the diffusion rate curve, '.Io express the equa tion in te.l1ns of K, -dw cf~ =-·---! (P) L All the quantities involved are constar..t £or a given experiment or are readily obtainable from a plot of the time rate of increase in wei~t of the drying agent above the diaphragm,. '.lhe experimental procedure is to provide a source of constant humidity for the vapor spe.ce and to let equilibrium conditions f'or the initial aqueous tension to be established before. the s-t;art of' the variable pressure gradient experiment. u11h en such equilibri~, as shown by a constant rate of diffusion, is reached, the supply of moisture is removed, and the vapor pressure from then on is allowed to vary with diffusion. Ept~r~ntal Results, .,. 1he method used was refined as far as was pre.ctice.1 6 but because of' the .nature of the materials a great deal ef me.nipulation and a 6 ree.t deal of time were required to obtain a body of reliable data. It was necessary to prepare each test . i 1____ _1_:_ __ . I ➔·- t-----·- l - _j -- ·-- • I -·i--! 1 l • ---t--- f- 1-·-· · 1 I ..... - ' ---L-- ' ·I i .i .. ~ ··· -1-·--I '- -- i .. l1 .. - - --r, I I _ L_ i----- - 1-- r ··:- - ··i- . ' I . -t~ ~~tt~A11t44~tFt~mtrt+qr ....- -~--y•• •• .. iI . . :~_:_ .. . : -- 1 - .. ·· 1-. . -1:-- __ I - --~- - - - __, _ .. . . .. . , . . -1 -- -- - l i i- -+--- ~ •• · - ., J __ •• -- - . ..... ---· . -· .. -~-------- i ! ! ! - ---, -- - +:·-- --~- ~- ----1-:--- :-- I • : • ' F: ··: :. -· :· --- ----- -- ~:-:. ··•·· 1- · ·· ·"· ·-1- •••• __ :__ 1 .,! i ! . I :. ·-·· ·1· ··-·· _____ :..__ i . . _ -- • ---1---- I i ·--t-· ---i-·-- · :·:- 1-- · · ••••. + ·-··- -+-. - - . . . . .. - - - - , - · - ·· - . -- . ·• ! _· - - ! I - ---··--, --L!)-i- . -. ~ : _·-t--i-:-:--t-+-+-+--+--~ r. -r--~-t--,--l---+-+-l----..ii1,..1- - ~ - --~ ····- ~ ·1 ·~---· --· 7 - ·-· i . - - - .:.: .:... j ~ -- --1 -. l~- -- -r--:--:-n~~~-:t~~r.+~~+-4;....:...+--1--~-~-~i ~ ~~I T~n H-dd --i-d~ ~ ~h-f.:.-++ +++~~1..-Ll .1!~~- __:·i -_ ---~ -i-- -----1I•;t: T ,t : -~ .rI ._ _ i . I ·r---t r ::t-f f + +--Prt4~q~_:_;_:-4_~.:~:~.-_;-__L_-__~ r---~--:- . ' -r- ~ r -r~ -:r4--:--~ r -t-+ ~ 7-:--:-~+:-+_µri:~~-~~~--L•~-l -_;~J_.:.- i-i-~[• -1 _j -·i -- t- - -T ' .. ' --- ·+ - ~ -.....a. j _ _ I r t:::t:.T t f f --t.:. :.+--1~ 4 ~hh +--*+-+-+-i-i--1i ~ i +-·- I -1- +----+-- ' ! -- .l ·- - ., 1 --- --f------+-+---i-~ 1 1 i •i !• · ----~- ----- ---:--i • i - <-~-- . . . --+---'. -~ .:. -~i ·-+-----+- ·I I "' - 2 - %r~:7:"t---t7 ~ -±7~++ +--+0.~~k~LJ~_j~d "' diaphrag1n and to measure pertine nt dimensions on eaeh individually. '.1'1:te l ength of t i me needed for a single experiment to p:r·ooeed to t.l-ie po int a t whi ch an accurate diffus.ion :rate could be de termined va r ied from 150 to 1500 hours., depending on the nature of the materia l. From 12 to 20 observf;l.tions were made on each diffusion cell du ri.:a$ the course of ea.oh experiment, in order to obtain a .s ufficient mtmber of experimental points to de fine with good precision t.he dif.'fusion rate curves. Each run was me.de at least in triplicate to dim.inish possibilities of error. A total of over 6000 ob- servations haVY:) been made dl.U'ing the work on diffusion, 'Ihe qua.rt, spring be.la.nee diffusion cells were found to yield quite satisfactory data, Typical curves ehowing experimental re. \ sults a re given in .Figures 9 and 10., where the centers of' the circles represent experimental points and the size of the oirelea represent the proba ble error of the n1eaauroment. .A siniila.r curve was drawn 'f or each membrane,, the slope of the curve giving t he diffusion constant. Copsta.nts at 98% Humiditl (l:t; 25"C.,; • 'lhe gre,a te r portion of the diffusion meaeurements Wfpre made with a constant partial pressure gradient of 23.5 mm. Hg. are given below in Table IU. '.!he results of these experiments 1he !'ii;ures recorded are the dif... gms ~ ....,..., • tor- each fusion constants, expressed in the units of __.,.,...,...,.._ om. mm. hr• individufll diaphragm. Table !II Mat erial NQ• M.,P. Pen. Diffusion Constant x 109 Average Constant x 109 1 174 5 4.02 4.00 s. 9'7 2 165 2 4.06 4.05 4.04 3 204 0 ~3100-4 159 15 5,92 5.87 s.12 6.03 o.oa D•3100-6 126 45 s.2, 9.,28 4.82 4.?3 4.76 234 l 9.27 M.aterie.l Wo, M.P, Pen, Diffusion Constant :x 109 Average Constant x 109 E•519S..2 140 25 7.54 1.57 1.sa 7.63 M..l 190 7 5.40 5.,38 5.39 5.. 45 5.33 176 12 168 20 6,92 6. 72 6,93 7,14 142 34 ,,_ 74 7. 70 7,98 11 166 24 9,35 9,43 9.,20 9,20 9.so M:At erl.al No. M.P. pen. Diffusfon Constant x 109 Average ConiJtant x 109 12 188 19 D- 3100-1 169 20 8.27 8.15 8.25 D-3100-2 190 20 9.25 D-3100-3 215 18 9.25 21 210 19 a.so a.1s s. 76 a.a9 22 244 3 s.s6 3. 6'7 31 140 150 24 3.• 32 a.'Tt ildaterial Mo. M,P1 Pen, Diffusion Oonste.:nt :x 109 Av~rage Constant x 109 32 160 ·• 2 2.16 2.15 170 2,.14 2.12 180 ... 1.4 2 190 41 180 ... 2 195 o.434 o.435 o.437 The materials tor which only one £igure is given a.:t'e extremely brit·ble at ordi:rmry temperatures,. and therefore very difficult to work with. The other two of the three diaphragms - and, indeed, others prepared• fractured. However, it ia reasonably certain that the siri..gle values a.re n(;)arly correct. Dtf.fuaion Constants at Various Hurnidi titHH • Dif.tusion constants were evaluated for a number of' the je.mples at different constant aqueous tension gradients. 'lhe resul ta obtained a.re recorded in Table IV as average values or the triplicate runs. 'fue agree'!llent among the individual values is much the same as that in Table .Ill, Table IV Vapor pressure -, mm.. 2s.s 18,.8 11,9 7.,31 Me:terial No-. Nt.. P, Pen l 174 5 4.00 3.8S 2 165 2 4.06 3-.82 3 204 0 3,07 3.• 02 D-.3100-4 159 15 6,00 a.se 0-3100-6 1a9 "I 4.71 4.93 s.o4 11 l66 24 ~-~30 9.00 8-,55 at 16,9 mm. 12 1.88 19 a.• 16 7-.&5 7,50 21 210 19 a.79 1.,10 22 244 3 3.66 z.26 D,,,,3100-1 169 20 8,2.2 a.oa D.,.3100.2 190 20 9.,25 a.99 Sl 148 - 24 3. 37 s.a2 4,oa 166 24 9,23 8,.93 8,23 6,7(:) a.03 'l,85 s.. 40 150 * ll * The.ae values were obtained from the variable-pressure :method described on Di~fusion with. Water in Contact; "' 3ince in practice liquid water is often in contact With aaphal t, it was thought im.por·cant to '.:~establish : whether the diffusion constant would. be much different if liquid water vrore in conta.ct with the diaphragm instead of water va.po:r_, '.Fne results, given below, indicate that diff'usio.n for liquid water i~ definitel,Y faster, e:speoil.ally for sceam.. re:t.'1:ned asphalts. Te.ble V K to'i:' liquid l 5,15 2 5.10 K for vapor at 23.3 mm. 5.15 9.,30 ll 12 21 s. 66 22 10.00 In ord!; ) r t o a r rive a.t some conception of t.."".e magnitude of change in diffusion constants with temperature, some expe:riinent3 ware conducted at 35° c., using saturated NH 4 Cl to regulate the humidity, The results a.re compared below to constants for the sam.e asphalts at 2.5°G. under equal relative humidities. (ao.l.%) l 2 7.55 ll 1.65 ,12 10.10 Dif1-U$ion ·throu~ PiJ?! Line C~~ti;:i,* .. As a ro.atter o;f' praetic.a l interest, several portions ot n pipe-.line coating which had. bee-n in service were obtained. nits m.~terial had beon on fl 12-ineh hot line at the Los Angeles re.finery of the Union Oil Company since 1927, '.fe13t specimens were used for the diffusion runs in tho condition in v,·hich they "'<'Jere> received, Values of K for several test:; did not cheek, but all were in the region of 3 x 10-3 . . gm h era.mm. r. The date. reported, with the exeeption of 'those for i;he pipe .. line coating, tu-e for i'reshly pt·epared asphalt. Atte: ,1pts ·,:~ere made te d~vise a method of weather±ng th.in diaphragms 1n a W~ather-o....t.:eter and subsequently :meo.surilll; difi'usicm rate$; but, the tempel''f;;.tures reached durin5 the artificial weutheri:ug mad1;.; it thus far e::ct,r6W:)ly difficult to devise n support for the aspha,lts in such a fashion that diffusion experiments could later be run with reasonable certainty. l)iscuasion of' Results; .. From. the dat a obtained a number of interest... mg obs:orva:tion~ rnay be made concerning the dependence of the diffusion eonotants on various fa.otors. 'D'J.X: of Mnt!1'ri.al. In general., air.blown asphalts offer the lea.st r esist ance to ·the pa ssage of.' wate r , a:nd eoal tar pi tohes the most. Const ants for stemn,... refined asphalts a. re lower than those for -th0 air....blcn!Jll t ype. and higher than the constants t'or eoa.l ta..r li'il;.ler,. Diffusion through t he filled products ava ilable vm.s. found to be slower than through an unfilled ma.terio.l correspond- ing to t hat from whioh the enamels were ma.de. However,- since only three materials containinn; mineral .fillers were at hand• no generali• ties may yet be drawn. ~e!r!l~e, The dU'f'usion eonsto.nt inoreases with tempera• V;Laeos,tty. In an attempt to find correlations bet"1'reen di£.,. ' tusion .ra:tes and other properties of the materials studied., plot& • were made of K versus penetrat ion for ea.eh type of bitUU1en. For the case of the ab·...blown asphalts there appeared to be no r egularity to the values of Ki bu.'c for the steam.refined asphalts the plot gave a eurve of exponential form. 'linen the de.t a were plotted on a log• lof; sea.le, a st,raight line resulted for ea.eh series of stew.n•blown a.sphal t.s, li'i gure 11 is suoh a plot of: lo.g K versus log Pen. The relationship between difi'uGio-n conata.nt a.n d penetration fo.r ·t;he 13eriea of stea.m..-blown asphalts from crude source II A11 is described by the equation lo~ K + log 109 ;:; 0.33 + o.59 log Pen Figure /I Corre/or/on o/" Oir/"usion ro Penerra.rion. 0. 7 0.61---+:7"""'-'F----'---+--+ >l . 0.5 ___;____L___ • 0. 4 I . iI --•· - 1 - ·•- i . =r- +---+~+---4--+--- + -+----l--+---+--+---+--+---l- ., I ·- _____L __ _ i i ···+·· ·- · ,-l \ ... J ___ . ·-1... I .; -+----+-+----+·-+---r--• I ! ! ·--r--- ! t--- -- ..' ! TI - i and £or the asphalts st eam distilled from Pa.nuoo crude. 9 log K + log lO = o. 52 + o.25 log Pen.. A fundamental relationship(a) between the pe1.,etr-ation and the viscosity of asphalts has been d,: : riveci and generally a.eoepted by workers in the field. The expression is 5. ).3 X 109 -= Pen ~ ' i 95 , poises ~ ., Y1 which holds for asphalt~ 'Which a.re o:f' the na:cu:re of Newtonian liquids; in that they e:ltilibit visool.lS rather than ple.stie :ti.ow. Since coal te.r pitohe$ and stee:m.-refined asphalts t:i.pproaeh lfewtonian beha:vior( 9 ) 1: we are justified i.n oonsidering this equation for the oase at hand. Taking the logs of both sides of the above equation vre have log~ =log (5,l x 109 ) • l.93 log Pen. It is readily app&.rent that if the diffusion constants are e. linear £unction of the penetration. they are also at least an approximately linear function of the absolute V'i~cosity. In order t0 disouss mere freely the properties of e.sphal ta, something mu$t be sai,4 ot their eomposition and structure,. Asphalts are generally considered to be a eollotdal disperaion o.t very hig,b. molecular weight hydl'eoa.rbons of an aroma.tie OT hydre-a romatio nature 9 in a mixture of resinous materials and oila. ( ) If the dispersing phase is highly arome:t;io, the dispersion will be well peptized, and . the viscosity will be e. i'u11ction of the oily phase and of' the :relative ameunts of the two pha.ses, If tho a:mo-.mt of aromatic oompoun1s is reduced., as by air.. refining, wherein the aromatics become polymerized and become t ransferred to the dispersed phase, the peptbing ability of the oil i'r.e .otion is reduced, and plastic flow che.racteristios a.re developed. On the other hand., ii' stea.-n-distilling is carried far enough, a point will be reached where an inversion o:t' phase-s result&., (lO) end as that is approached the flow characteristics of the material are no longer sim.ply related to those of the oil and resin fraotion. I.f these statements are asswned to be essentially .fact, it ma.y 'be deduced that d:i.:ffusion through stear.o.... refined asphalts is dependent ori the pl"operties of the dispersing phase» and that the prineipa.l efi'ect of the colloida,l micelles is to hinder the transport of the ws.ter molecules through the oily fraction. A condition im,.. posed. upon thb, ho....rever, 1a that it applies only up to the point at whioh the su pposed inversion of phases takes place. '.!his oondition i1:1 stren,;thened by the fact that the hardest a&phal t from crude sou:rce t'A'' yielded e. diffusion eo:nstant {Penetration 0) which on the log-log plot was quite far from the straight line through the other .five points. The t·wo curves relating diffusion eonatants to penetration he.ve different slopes by a substantial amo1mt. '.lhe o'b$e:r'\1'a:tion might 'be made that the two series of asphe.lts are quite different in temperature susoeptibil.i ty, which is a .tun:etion. in the last analysis, of t..11.e chemical na.ture of the constituents. Vapor Pressure Gradient, • lhe de.ta presented for the variation or diffusion rates vrl;bh humidity g:radients indicates fairly conclusively that there is notiQeable digression from linear diffusion as represented by Fiek•s eqUation. This sug 6ests the presenoe of water-soluble materialS. or of adsorption effects. Sufficient information i .G no'b y$t at he.nd ft-om which to deduce the cause of deviation i'rou1 s.i mple di!'fttijion. However, to indicate directions of a.tte.ok 1 it migh"t; be $t:at&d that if the distortion of the prenure grndient is due to _the solution of wa·t er,, it is probable that diffusion coni;.ta.nts measured under condition.fl of humidity l'IJhich are below those for saturated solut.ions of the hygroacopie ma.ter:ie.le will be independent of t.he vapor pressure in the range below thnt critical humidity. Eut if the vapor pressures or t;hose solutions a.re l<>w., the re~ults will be experimentally indbting:uishable from those expected for adsorpt.f on on surfaces. v,·hich would vary wi.th vapor concentrations along a cunre like an adsorptio.n isotherm. An investigation of the o-on... centratio.n g.: radient by means of lruninated diavhragm may shed some light on the problem. Mechanism of Diffus~on; • The concept of' linear diffusion implies the transport of water as vapor throug.h the interstices in the solid by means of molecular motion.< 7 ) Conoent:rations of ·water vapor within the solid would then be somewhat less than the eoncentration i.a the adjacent vapor phase because of the spaoe ocoupied by the solid, but would be directly proportional to the concentration gradient. For diffusio-ne.l processes cilaraotarited by this mechanism Fick 1 s law a.ppliEHI unrestrictedly • .A8phalta contain a number of polar molecules, suoh as oxygen and sulphur oompounda., and it would be expected that these exhibit a certain affinity .for we.tar. fusion process in s.evere.l 1.·mys .• This may eonoeive.bly affect the d:i.i'• When 1.l'Je.ter vapor is adsorbed. energy is libe:ratea~ and the fugaetty is lowe:red. Vapor then moves in to tal.-re the plae0 of that adsor-bed., ru1d possibly a second layer beeo.mes associated with the fit'Ef'c layer a.dsorbed. lheae a,lceorhed la~rs 'f/JIJ,Y be subjeet to strong apreading foroesj> thereby providing a means o:f transport which me:y be muoh ,uore rapid than by 11'..oleeulnr motion througJi 5,nter•moleoula.r space.s , . 1\ir..-blown asphalts give evidence of str11eture(ll) • while s t erun.~refined asphalts and coal t@ir pi tohes are nearly homogeneous,. .Movement; of adsorbed wa-te r alon.E; the struetur,e would explo.in. the :rea~on why dif'i'usioll through the air-blown. type is more rapid than th:r·ough the others., 'f!t' hen s:. moleetlll!it is adsorbed. ene1·gy in sub.s te.ntial runount is liberated; Em.cl. likerwise ernerg,y mu.st be supplied to remove a molecu.le from an a<.lsorben-t., The quantities of energy so involved depend upon the st ate from which the molecule was adsor·bed or the state to which it is returned• the hee.t requirement i ucreasin~ with a de crease in pressure. Then a 1.'fato:r molecule would be ad.a.orbed from a high humidity a t a low· heat of oondensation, a.nd would, be :rem..oved at low hu..iidi ties only by the absorption of a. higher heat of vaporization, Under ioothermal conditions this would result in :retarding escape from the low pressure side of the di.taphragin, and w-ould caust~ a shift of' 'the coneent:re.tion towtird the low.pressure eide until ~ energy ball:.>.ll.ce was obtained. The effect would be to change the rate of di££usion with concentration g:radient. w*;,tor i s 2~n i::;.1_p()rtant factor, i't im of value to be able to oalculate how muah ei£ e. giv·a n protective ageut m.ust ba use d. i'o~ provision of the most ei'fieicnt pl--otection under a given set of conditions, .Although each ;p:roble.m poss~sscs its own peoulia ri ties,. nav,n·t.helesa a i;enere.l type o:f.' upproaeh may be in:lioated,. !he only condition :required for the transport of vmter through this diff':arcn•l;is.l is produced by a dry sp&ae on one side; by solution, by· adsorption, or by chemical reaction .. such ~s the co:..~rosi.011 of a metal pi!)6 lino, is ci' small ccnseqmmce to the m.echanic:rn oi' the o&.luulation. transferred int.o a spac-e iui tially r1~ee of water., the pressure of '1$."ter vapor in the space being built u.p during the process • .tit any instant the pressure, P2 ., in the space will be re.. presented by where W :::: 'mp:l,ght of Ymter present in the space = weight of water diffused through the ma.teri.al V = volume of space c :::. conversion faotor bet·ween vapor density cmd presm.:u.·e4 From thl:) linear diffusion law, which will apply quite closely it proper diffusion cons·bant.s are known,. KAt w:::: · L (-P *l .. P) ·-2 · A = area of the dii'i'u.sion path For an in.crenwn-t of t :L!lle dt ksaund.ng it is desired to know tb.a length of time required for t he transport of e. t;i•:r1:1n amount of water to diffuse through a g;i"1ien t hicknt,ss of coe.ti11g,, the derivation is as follows; C Since P2 :;::: V '{f v aw C d.Pa :::; Also d.P2 = --d(P1 • P2) and V dYf =: - d:?2 0 Su'bsti tuting, dt = VL KAc VL dt=•K. A c VL K AC Integrating• d(Pl ... Pa) Pi - P:;a VL e ~.-: ~ K .A. c The:n t 1n l'i = ... _JI,,. KA e ' VL -t=KAo VL KAo t VL e P1 ... P2 K .Ao t (Pl .. p,.,) .., e VL = Pl At e:ny given time V .:: c ---- --l=• pl V W 0 ... log ( Pl V = ... e Pl (l • VL ) K *.A. o ...... t ii' C and t KA ct - e VL 1) ... YL 211 303 K .A e ::: 2.303 105 ( WC P V 1 KA c t ''IL •l) From. this expression rn.ay be eale-ula:ted t.rie int;er... relation of t and w, or of W a.rid L at constant t. li'or thin membranes of aa phal t the diffusi.on constant i.rl.ll not vary appreciably vnth change in pressure differential, so i n that case the equation may be solved formally. If thick layers greater than a.bout o. 25 mrn.. in thiekneas are to be used, variations in K with the quantity (P1 • P2 ) may be taken into account if knowledge of the specific oondi tions is sufficiently preois& t.o warrant it., As a specific ease for illustrating the calculation• the length of time required fer a layer of water molecules two deep to be built up on the surface of a hydrophilic rock covered with asphalt will be estimated. Assuming a. -;.~.t~r molecule to be approxi• mately an oblate spheroid of maximum. and minimum. :radii of land 2 Angstroms., respectively, consider the time :required if th.at aro.ount 0£ water on the rock has a fugaci ty corresponding to 16 mm. Hg. The volume term, for the basis of estimation, may be considered to be that which will contain that amount of water as a vapor at a vapor pressure of 16 mm. '.Ibis assumption will result in a longer calculated time than that actually obtaini~ it' adsorption occurs, but will give an approximation muoh better than a conjecture • .14;grega.tes in a hot mix often have a coating of asphalt only 5 microns thick. Assume for convenience a thickness of 100 mioren&. Fer an au-•blown asphalt having a constant, K, of 10- 8•~ the calculated time is 2 3 {2. 36) (10· 2(10· ~ (2. 303) (10- 8 )(1) (106 } t =1.es x 10-4 hours, tor a case in which Pi is 23.76 mm... Hg •. Similar ealeulations could be me.de ro.r other problems, such as the penetration of moi~ture into the organic fibers used in asphalt roofings., the eo:rresion rate of protected metals, and the like. If conditions are known, the c0mputations will be an indication of the amount of prowetion again.et water which will be afforded 'by the material in questio.n., Surface .and Interf'a.c ial Tensions In a great many applications of asphalts as protective coat• inge., displacement of the asphalt by water oocurlJ or is suspected w have occurred. As an instance, when coated pipe•lines corrode at a place where a ere.ck in the coating has appeared, the corrosion it not limited to the region exposed by the rupture, but is quite general on either side under an apparently sound proteetive layer. Otten deep pi ts are found some d:tstanee £rom the oraolt. A possible reason tor this may be tru,.t water displaoee the aaphalt. thereby forming e. channel alQng the surfa" ot the metal in which water may move freely, providing a. means by whieh e:ictensive corrosion :may take By measurements of interfaeial and surfa()(;) tensions the .free $urface energy change accompanying the exchange of an asph:al t-water and a :mQtal.,,•wn.ter interf'll\ee £or an a.spha.l t-motal interfaoe may be computed. The sign of this energy change will be on indication of the energ~tie possibility of' such an exchange of interfaces. When correlated to the reeul ts of other simple experiments, such as simple measurements of adhesion in the presence of water, informa. tion m.e.y be available from which to deduce ·whether the displacement takes place~ or, if it does, whether it goes .fre,ely or only under mechanical shooks and streoses. E~e.rim.ente.,l Metho4s, and Resul t~.i • A oonstant temperature bath. has been constructed and adopted for use with a Du NW>y tensiometer. Preliminary measurements of asphalts and pitehes at various temperatures above the softening poi.n t he.ve been ma.de. Extrapolation of the resulta to room temperature, however, is found to be inexact for most of the materials because o:f the length of the extrapolation, The data ta.ken indieate that at room temperature the surfaee tensions are approximately 40, 46 and 65 dynes pe~ centimeter for air.blown asphalts, steam,..blow.n asphalts, and coal tar pitches, respectively, An indirect 1JWtho4 of obtaining surface tensions of a$phalta is the measurement of surface tensions of solutions ot asphalts in ·benzene ~t lmown oonoentrations and extrapolating to 100( oon• centration. Typical results obtained are as follows, Concentration Surface Tension of asphalt Dynes per om. 10 % 2s.2 20 2a.4 30 2.8.• 6 4,0 2a.e 50 29.l 60 29.8 The 60% oeneentration represents practically the limit of the solubility in 'benzene. Al though this method has been used 'by other wor.kers to estimate surface tensions. it may be seen from :inspectian of the above fiaures that an extrapolation from 60% to l ~ is subject to a large proba.Qle error. 1he trend 0£ the figures indicate that beniene is probably capillary active with respect to asphalt, so the greatest change in su:r:fa.oe tension most likely 000un in the region of 100:i asphalt,. which is impossible to reali~e experimentally. It was concluded that for reasonably exact d,etenninations this method is without merit. The so-ftening points of the :materials e.t hand are too high to permit direct meaeurement of' inter:f'aoial tensions between asphalts and water;, so preliminary ~xperiments were eonduot,ed on the determine..ti9n of asphalt..Ym.ter interfaeial tendons by a. method involving benzene solutions a.s in the method described above. Since the inter.. :t'a.eial tensions v,ero found to decrease with e,n. increase in asphalt concentration of the benzene phase, it was decided th~t thia pro ... cedu:-e me.y be of value. Typical results are shown belowllt Concentration Qf asphalt Interfa.eial Ten$ion Dynes ;pe :r cm., 0% 31.6 10 23.1 25 22.9 45 21.3 66 20.1 Chromat ographic Anags:j.s Adisorption methods have been used to remove resins in on~ step of' a Selective solution procedure for splitting asphalts into three fractionsi but no known attempt has been made to fractionate e.sphe.lts entirely by adsorption methods. 'When a mixt7.tre of molecules of different ldnd,s and sizes are passed through an active adsorbent. the moleeulea are sep.u-ated.. aom.e-td.mes quite sharply, a.~oording to chemical nature and size. In many ca.sos the separation can be effected quantitative.ly if' the a.dsorbing material has the proper activity-. and the solvent chosen for the miJCture to be separated is not adsorbed rnore stron$1Y than the aompouruis w:hieh are desired to be removed frolll the mixt~re .• Asphalts have thus far practically defied attempts a.t exhaustive organic analysis. However. even though the great number ot individual compounds present may not easily be identified,. it will be of consequence to a fuller und0rs-tanding of asphaltio produets ii' gl'OUl)S of similar compounds can be isolated and ehan1.cteri~ed as ·t;o amount and kind. 1he poasibili't;y was entertained that selective adsorption might otter a new and eon~nient met hod of fractionation of asphe.l ts. Experif1ental Method • .r~re..ctionatio;n by Adsorption, • l~ctive.ted alumina.. ground to 80 ... 100 mesh in a bs,ll•mill, was selected, sinee it gave the most promising results of any tried e.s the adsorbent to ·oe used in fractionation of asphalts. A glass column~ 16 inm. in di.runetet", is pa~ed t o a he i ght o.f 10 cm. with t he alumina, and a measured portion af a solut ion o.f asphalt of known 1t1eight concent ra<bion in XylE?ne allo,~red to run through under suotion. The column is the n 1.\-ashed vlith additional Xylene to remove ,my unadsorbed material. The pa.eking is then pushed out; cut a.pa.rt at the juncture be-t;v,~en be..nds of' different eolo:r" and av.ch band e.gi tc.ted for a minute or tw-o w1:bh benzene and a few drops o.f a lcohol., the t-i.lo@hol serving t o displa ce the substance adsorbed. 'l'o reduc~ the amount of solvent r equir'e d to wash t he alumina free of the displaced fraction., each pot'tion af' tht~ column pt1qki1~ i s extracted in s. Soxhlet extracto:r with the solvent mix.tu.re with which it had been agitat ed. The solut ions obtained from the e:x:tre.cti ons e.re each evaporated unde:r ve..euum in a 'tEu·ed f'le.sk, so that the weight of each fra.otion may be obtained. Actiinition of Alumina; ... The original e.lumina whieh was crushed in a ball-mill was used in preliminary work wi:thout any tree:ttne:nt to increa se i"t;s activity.. Spent alumina when re&.ctivated by heating proved to be so e.otive that no separation could be obtained with it until it he.d been thoroughly exposed to normal atmospheric conditions. When it had been so exposed, re~ults obtained with it checked very well t he original work. Oha.re,eterizatio;n of fractions: .,. At this st ~lge of' the in... vestigation molecula r weights of' the f'ractions are being deter..iined. lhe procedure followed is t h at devi.sed by Barger(lS)• '.!he principle on which i t is based ie thut if droplets 0£ two solution$ of un_equal vapol" pressure a:re placed in a eapillary I one will grow at the expense of the other., l'.his growth i .s observed by mean3 of a microscope. fitted with a mie:rometer eye•pieoe .. &lt d can oo detected quite r~adily wit hin a few minutes it' the t wo dropl®ta are much diffe r ent osmoti"" ea l ly. A S.olution of known weight concentration ot an unk..l'l:Own com,. pound may ree,dily be compared. to several solutions of known molal oonoent:ra.tion of' ~ kn.own co111pound.,and the moleoular w:eight oi' the unknown obtained by tri.e.l. 'lhe capillaries used are e.ppr0~ir.lately of the same siae as those used for melting point detennine:l..i.ons. Iilecnzene is used as a solvent both for the unknown and for azo• benzene., which is used as a reference compotmd. '.the droplets are placed in the capillary tubes by means of capillary pipettes. Efferimenta.l Result,s. E,r,actionation of asphalts: .,. 1he procedure developed effected a separation of asphalts into three distinct fFt<.ctions. The upper 4 cm, of the paoke4 eolumn beoame brown:hh•blaok; the lovre:r 6 em.. became a faint reddish color; and a solution having the appearance of e. highly refined lubricating oil v1ent oomplf.ltely through the coh:mm.. De.ta t'or sori.lEl of the asphe.1 ts are pre$cmte:d bel.ow-. E ...1 • 7.95 45.6 7.50 4s. e 47.S 7.8 45.9 47.6 6.6 45.• 7 99.. 8 49._5 s.o 44.4 100.5 48,5 4.5 45.4 98.4 Run with reactivated and conditioned alumina. These data e.re uniformly quite sa.tis!'actorJ♦ rt is thought thl:l.t the deviations are due more to JM.nipulative ~rrors rather than to some f'ailu1·e of the method, 1'9,lp~~l~ 'W>'eiW!~'• .... '.lhe technique of thia mioro<l!lm.ethod of molecular weight determination has not yet bean fully de veloped. 1 but work with the tv,o large fraetions of the a.aphal ts has yielded results whi¢h e.re felt to 'be exact within 5%.. By t:ria.1 the molecular weig..l\ts o:f' the two largest fraotions obtained from a.sphf\lt No. D-3100.-1 were found to be 1425 for the top fraction and 675 for the v.nadsorbed fraction. }lt.eouesi.on: ~ The method of selective adsorption as applied to asphalts e.pparently holds some promise,. .Additional fraotion!'J might possibly be detec·hod by viewina; t..lJ.e column unde r ul tra....violet li.ght, and re• fractionei.tion l'.tight pro:ve to effe.c t further separation., '.the exoellenoe of reeovery of material f':rom the uiumina indicates that there is some probability that th~ p:rocedui•e might be made guantitecitive., .. 54,... 1. .An apparatus for the accurate measurement of angles of oontaet be~veen asphalt surta.ees. and liquids was eonstruct~d • .Measurements show that the ease of wetting increases in the following order~ filled air-blown asphalts, unfilled e.ir... blown asphalts,. steam-blown asphalts., coal-tar pitches. The angles assumed by wa:ter in contaet with air•blown asphalts a.re greater than 90°, hence water does not tend to enter cra.oks in a. fresh air.blown asphalt surface. For the other materials the angle is less thall 90°., which indicates a ·bendeney for vF&ter to :penetrate under the action of surface .fo roes. Contact angles :f'or all the materials are redueed in highly alkaline selutions, the reduction being most pronounced in the ease of coal•tar pitches throughout the entire alkaline ral'}6e. '.Ihe effect of Ca.Cle solutions and of HCl solu·tions on contact angles is very· slie;ht, excepting that the wetting angles fo.r coal ... tar pitch are noticeably reduced 1n ECl solutions. 2. A cell for measuring rates of diffusion ,rl. th e. probable over..all 1 error of less than s. a:, was developed, .A method and an apparatus for the construetion of· quartz spring.a were devised. Measurements of' diffusion rates indicated that water transpoTt is most rapid through air-'blO\"llfi asphalts., less :ra,pid through steam• refined asphalts., and lea.st ra.pid through coal... te.r pitches. Diffusion eonstants were found to increase with temperature for all clas$eS of bitumens studied. Evidence that diffusion through these sub- stanees does not follow a simple diffusion law was discovered. A correlation between diffusion oonsttmts and visoosity tor steam...blown asphalts is presented. Applications of diffuaion rate data are indicated. s. Preliminary mea.suremen·t;s of sw:-.faoe tensions of asphe.1 ts and oj;.' intarfaoie.l tensions bet-,,•1oen aspha.l ts and 'lire.tel· \>Jere ri1ade. The method a_£ detennining the surface tension of asphalt at room temperatures by extrapolation fl'oro. rn.easurem.oots with beuz:ene solutio~ cf e.sphal ts. was found to oo invalid,. 4. A method of fractionating asphalts by chromatographic e.nalysis methods was developed, and evidence that the procedure might yield quantitative resulte was obtained, Biblio~raphY; 2.• s. Gaal, Worl,p. Petroleum qongresi, (1933) 9• Pf'ei:ffer and W...'rl Doormanl, tl, In:st. J?etr. Tech. 22, 414,..440 (1936} - Aeknowledgm~nt ~ . ' .. The author is indebted to the .Asphalt Institute for f'ine.naial assistance in this research., and wiS.hes to express his thanks to Dr. A, o. Beckman., whose aid and adviee hr.ve been invaluable; to Mr. Willis Routso..n, who assisted materially in the laboratory work; and to Miss Blea.nor Leeds for typing this thesis. Credit is due the Union Oil Company e..t Wilmington for the use of their Weather-o....Meter, For the friendly advice and encoure.gec,ment offered by Dr• van Doormaal a.nd Dr. Anderson of the Shell Development Company, and by Dr, Bray, Mr. Grebe. Mr, Scott, and Mr. Beekwith of the Union Oil Company the author expresses his gratitude.