46 TRANSPORTATON RESEARCH RECORD 1115 Size Exlusion Chromatography and Nulear Magneti Resonane Tehniques for Prediting Asphalt Yields and Visosities from Crude Oil Analyses P. M. BEAZLEY, L. E. HAWSEY, AND M. A. PLUMMER Size exlusion hromatography (SEC) and nulear magneti resonane (NMR) tehniques for prediting asphalt yield and visosity as funtions of rude oil type and manufaturing ondltlons are detailed in thls paper. SEC onditions were varied to yield rude oil omponent distributions to the asphalt phase that agreed with the thermodynami requirements for distillation proesses. Silanized-silla olumn paking and ultraviolet detetion yielded better results than polystyrene paking and refrative index detetion. Relatively standard NMR onditions were used to determine the ratio of aromati protons to paraffini protons. This ratio was used as a measure of the relative polarity of rude oll and asphalt omponents. Asphalt produers have historially relied on fixed domesti rude soures to yield onsistent quality. Reently, some produers have been required to use variable rude slates; this makes asphalt quality ontrol more diffiult. n response to this situation, mathematial models were developed to predit asphalt yield and visosity as funtions of rude oil type and manufaturing onditions ( 1 ). nputs to these models are analytial results that require only 2 to 4 hr to obtain. The purpose of this paper is to detail the size exlusion hromatography (SEC) and nulear magneti resonane (NMR) tehniques developed for these models. To predit asphalt yield, moleular weight or size distribution and polarity of the rude oil onstituents need to be known. The larger the onentration of the higher moleular weight asphalt omponents, the greater the asphalt yield over other onstituents. For a given moleular weight, more of the aromati ompounds than paraffini ompounds will preferentially frationate to the asphalt produt. The visosity of an asphalt depends on its moleular size distribution and on its aromati paraffini omposition. A more aromati asphalt with higher polarity or an asphalt with larger moleular sizes will exhibit higher visosities. To obtain data needed for yield and visosity preditions, SEC was seleted for size distribution and proton NMR was Marathon Oil Company, 74 South Broadway, Littleton, Colo. 8122. hosen for omponent polarity. To ahieve the most auray and flexibility in modeling, the size distribution and polarity parameters had to be measured as independently as possible. A typial analytial approah for obtaining moleular weight or size distribution data on a rude or asphalt is to first frationate the material. Then SEC and moleular weight data are obtained on the frations. From these results, a SEC retention time-versus-moleular weight alibration urve is established for the entire material (2, 3). This approah is quite time onsuming. Also, it is diffiult to determine if all moleular assoiation and adsorption fores have been eliminated The SEC approah used in this work was to analyze, in its entirety, a given rude oil and all of the various asphalts distilled from the rude oil. Then, the perentage of omponent distribution to the asphalt phase was alulated and plotted versus SEC retention time. These experimental distributionversus-retention time data were ompared with those expeted from thermodynami priniples for distillation proesses.the SEC proedures were varied until the atual omponent distribution-versus-retention time data agreed with the thermodynami requirement of a monotonially dereasing funtion (Figure 1). n this manner, SEC onditions were developed that assured a true omponent moleular size distribution essentially free of assoiation or adsorption problems. The typial NMR approah for obtaining aromatiparaffini (polarnonpolar) omponent results is analysis by both arbon-13 and proton NMR (4-6). Then various arbonhydrogen ratios are alulated to obtain a measure of the aromati paraffini omposition. This approah is quite time onsuming. Only proton NMR was used in this study beause it was found that the polarity interations that ontrol asphalt yield and visosity ould be modeled with the aromatiparaffini proton ratio. EXPERMENTAL PROCEDURE The following SEC and NMR tehniques were used to obtain moleular size and aromatiparaffini breakdowns on 5 rude oils and 11 asphalts distilled from these rudes. Using the results, asphalt yields and visosities were orrelated to distillation onditions at average auray levels of.6 perent (1).
Beazley el al. 47 :;..Cl Q) ~ (J - ClJ. ~..:: OCl. ~ Q).<: a. "' C.<( E a> O.r: Ut- -o a; t- (.) Q; Cl. naurate Shape, < Due To Moleular ' < Assoiations 1 -- 8 6 4 2...s naurate Shape Due To Adsorption Loss On SEC Column Corret _s Thermodynami Shape-ie, a Monotonially Dereasing Funtion norret Shape Due To Detetor Sensistiviy ~,, 1-'-~~~~~~~~~~~~~~~~~~~~ nreasing Dereasing Moleular Weight FGURE 1 Thermodynami requirement for omponent distributions to asphalt phase versus SEC retention time. SEC The liquid hromatograph onsisted of a Spetra-Physis Model SP87 pump, a Spetra-Physis Model SP84 UV VS variable wavelength detetor, a Spetra-Physis Model 748C olwnn oven, a Basom-Turner Model 412 data aquisition system, a Valeo Model HH-6 injetion valve, and a Phase Separation LTD flow meter. The hromatograph was ontrolled with an in-house-designed miroproessor system based on a Motorola 68 CPU and a Pro-Log STD BUS. A silanized 6-A silia-based (6.2-mm inside diameter x 25 m) DuPont Zorbax PSM 6S SEC olumn was used with a Brownlee (4.6-mm inside diameter x 3 m) RP-18 guard olumn. The olumn was eluted with helium-purged tetrahydrofuran (THF, Burdik and Jakson) at a flow rate of 1±.1 mlmin. Best results were obtained with the olumn heated to 9 C and the detetor set at a wavelength of 22 nm. The detetor was not thermostated. During eah 1-min run, the data system aquired 1, absorbane measurements. Solutions (25 µl) of.5 perent by weight of rude or asphalt in THF were injeted. Both valve and olumn were thermostated in the olumn oven. Samples were prepared with helium-purged THF and stored in nitrogen in septum vials. NMR A Varian EM-39 9-MHz spetrometer was used in this study. Asphalt and rude oil samples were dissolved in arbon tetrahloride (spetrophotometri grade) to obtain low-visosity solutions. Tetramethylsilan was used as the referene standard. Other NMR operating parameters are as follows: RF power:.3 mg Filter time onstant:.5 se Sweep time Spetrum: 5 min ntegrals: 1 min Sweep width: 1 ppm End of sweep: ppm Sample spin rate: 4-6 rps Sample temperature: 34 C To obtain relative polarity values, two regions of the NMR spetrum were integrated. Paraffini protons were assigned to the to 5.33 ppm region and aromati protons to the 5.33 to 9. ppm region. DSCUSSON OF RESULTS As previously mentioned, the models need independent measurements of polarity and moleular size distribution. Development of NMR tehniques for determining the polarity input was relatively straightforward. However, onsiderable diffiulty, whih has also been noted by others (7, 8), was enountered with the moleular size distribution analysis. Gel permeation hromatography or, to use the modern term, size exlusion hromatography (SEC) was the tehnique of hoie. The first attempts used Waters µ-styragel olumns, THF solvent, and refrative index (Rn detetion. The data obtained from this system ould not be used. The apparent moleular size distribution of various distillation frations did not agree with the distributions predited from thermodynami priniples. The i;noderately high polarity of the Waters polystyrene olumn paking gave separations based on polarity and absorption in addition to moleular size. Also, the Rl detetor emphasized the polarity separation mehanism. For example, the hromatograph of a typial rude oil obtained under these onditions learly shows the effets of absorption and R detetor re-
48 3 -min F1GURE 2 Chromatogram of rude oil. Obtained with a set of two Waters 5-A µ-styragel olumns at 5 C eluted with THF at 1 mlmln; Waters R detetor at XS sensitivity; and 1-µL sample of 2 perent by weight. sponse (Figure 2). n partiular, it is believed that the separated Frations and are not olletions of material with similar moleular size but rather represent material with similar polarity. The large negative peaks, V and V, rendered the hromatogram useless for this analysis beause meaningful omponent distributions to the asphalt phase ould not be alulated. n an effort to improve this situation, an ultraviolet (UV) detetor was substituted for the R detetor. Using the new detetor, the hromatograms of the same rude and of an asphalt derived from it were signifiantly improved by the elimination of the negative peaks (Figure 3). Meaningful omponent distributions ould then be alulated from this plot, but the result would not aurately orrespond to moleular size beause multiple peaks were still observed. Then silanized-silia Zorbax PSM 6S olumns were used. The Zorbax olumns gave smooth hromatograms with no separated peaks or shoulders (Figure 4). These hromatograms ll V TRANSPORTATON RESEARCH RECORD 1115,: t _? A l <B 1,.,."'... 4 se 12 se -se FGURE 4 Chromatogram of (A) whole rude oil and (B) asphalt distilled from this rude. Obtained with three Zorbax PSM 6S olumns at 5 C eluted with THF at 1 mlmin; UV detetion at 254 nm; and 1-µL samples of.2 perent by weight in THF. appeared to be reasonable for moleular size distributions of omplex materials like rude oil and asphalts. t was suspeted that both the styrene- and the silia-based olumns ould absorb polar materials from rude oils or asphalts. Hene, hromatographi responses from both types of olumns were ompared. The same amount of sample was run on eah olumn under the same hromatographi onditions. The hromatographi results show that reovery from the silanized-silia olumn was muh greater than reovery from the polystyrene olumn (Table 1). TABLE 1 RECOVERY DATA Column Zorbax PSM 6S µ-styragel 13 A Total Area Crude 1798 92 Asphalt 4556 3223 - min FJGURE 3 Chromatogram or (A) whole rude oil and (B) asphalt distilled from thjs rude. Obtained with two 5-A µ-styragel olumns at 5 C eluted with THF at 2 mlmln; UV detetion at 254 nm; and 1-µL samples of 2 perent by weight n THF. To test the resolution of the silanized-silia olumn, two asphalts distilled in a standard refinery proess from a Wyoming rude were analyzed. The normalized hromatograms of these asphalts ompared with that of the rude learly reflet the expeted different omponent distributions (Figure 5). Also, a rude and its overhead and residuum frations obtained by the ASTM D 116 distillation proedure were run (Figure 6). To see if the hromatographi resolution orresponded to the distillation, the normalized residuum urve, multiplied by its weight fration from the whole rude, was subtrated from the normalized whole rude oil urve. The normalized differene urve was found to be equivalent to the atual overhead urve (Figure 7). Beause these urves were essentially idential, it was thought that the SEC separation with the silanizedsilia olumn orresponded to the distillation. This initial work was performed using a detetor setting of 254 nm, a olumn temperature of 5 C, and a sample onentration of.2 perent by weight in THF. These onditions
Beazley el al. 49 A FGURE 5 Chromatogram of (A) a Wyoming rude ou, (B) 2-25 pen, and (C) 85-1 pen asphalts dlstuled from this rude. Obtained with a Zorbax P8M 68 olumn at 5 C eluted with THF at 1 mlmln; UV detetion at 254 nm; and 1-µL samples or.2 perent by weight.,f A==.:.l' i!! J,_,, FGURE 6 Chromatogram of (A) rude oil, (B) overhead fration, and (C) residuum from A8TM D 116 distillation. Obtained with a Zorbax P8M 68 olumn at 5 C eluted with THF at 1 mlmln; UV detetion at 254 nm; and 1-µL samples of.2 perent by weight. 1 t-.q :; 8. Q) en.~~ oa.. ;:: 6 ~~ g_ ~ 5 ~ 4 o~ Q) (.) 2 Qi a.. FGURE 7 (A) Chromatogram of overhead from ASTM D 116 distillation ot rude. Obtained on a Zorbax PSM 6S olumn at 5 C eluted with THF at 1 mlmin; UV detetion at 254 nm; and 1-µL sample of.2 perent by weight n THF. (B) Calulated hromatogram from whole rude hromatogram urve minus that or the residuum fration. a a ' ' ~ norret Thermodynami norret Shape At 26 nm Thermodynami Shape ndiating Some Adsorption On SEC Column At5 C a Corret 1 Thermodynami~ Shape At 1 22 nm 1 b O +-~~~~~~.--~~~~.,.. -Q ' 4 5 6 -se FGURE 8 Distribution of omponents from a Wyoming rude lo a 85-1 pen asphalt versus UV detetor wavelength. Obtained on a Zorbax PSM 6S olumn at 5 C eluted with THF at 1 mlmln and 25-µL sample of.5 perent by weight. were aurate enough for seletion of olumn and detetor types. However, in tenns of the thermodynami model, the best SEC results were obtained at 22-nm wavelength, 9 C olumn temperature, and.5 perent by weight sample onentration in TF. n the thermodynami analysis omponent distributions to the asphalt phase were determined as a funtion of SEC retention time or moleular size. A typial analysis showed that dereasing the UV wavelength from 26 to 22 nm yielded the orret omponent distributions at high retention times (Figure 8). At the higher wavelength, small moleular size omponents were overemphasized. Another study showed that inreasing olumn temperature from 5 C to 9 C yielded the orret thermodynami shape at low retention times (Figure 9). At 9 C, the adsorption of large moleular size omponents on the SEC olumn was essentially eliminated. Using the thermodynami analysis, UV wavelengths of 21 to 28 run, olumn temperatures of 25 C to 13 C, and sample onentrations in THF of.2 to 2. perent by weight were evaluated. n these evaluations, the asphalts used were manufatured by a standard refinery distillation proess. CONCLUSONS SEC and NMR onditions have been developed for use in prediting asphalt yield and visosity as funtions of rude oil type and manufaturing onditions. The SEC onditions were optimized by omparing experimental rude oil omponent
5 TRANSPORTATON RESEARCH RECORD 1115 Corret Thermodynami Shape At 9 C - Q ~. Ql ~ V> - Cl).~ o.. :=:'. Ql ~ a. () O.<{ E Ql Or () Ql u (i; Cl. 1 8 6 4 2 norret " ThArmnrlyniliTii: ' Shape At 5 C Shows Adsorption On SEC Column Corret Thermodynami Shapes At 22 nm 34 4 5 -se. FGURE 9 Distribution of omponents from a heavy Arabian rude to an AC-2 asphalt versus olumn temperature. Obtained on a Zorbax PSM 6$ olumn eluted with THF at 1 mlmin; UV detetion at 22 nm; and 25-µL samples of.5 perent by weight. 6 distributions to the asphalt phase with the thermodynami requirement for distillation proesses. Aromatiparaffini proton ratios from the NMR results were developed to measure the relative polarity of the rude oil and asphalt omponents. REFERENCES 1. M. A. Plummer and C. C. Zimmerman. Asphalt Quality and Yield Preditions from Crude Oil Analyses. Pro., Assoiation of Asphalt Paving Tehnologists, Vol. 53, 1984, pp. 138-159. 2. D. Drul. Calibration Curve for GPC Analysis of Asphalts. Prt:print Papers, Division of Fuel Chemistry, Amerian Chemial Soiety, Vol. 26, No. 2, 1981, pp. 28-37. 3. H. Reerink and J. Lijzenga. Gel-Permeation Chromatography Curve for Asphaltenes and Bituminous Resins. Analytial Chemistry, Vol. 47, No. 13, 1975, pp. 216-2167. 4. S. Gillet, P. Rubini, J. J. Delpueh, J. C. Esalier, and P. Valentin. Quantitative Carbon-13 and Proton Nulear Magneti Resonane Spetrosopy of Crude Oil and Petroleum Produts, Part 2: Average Struture Parameters of Representative Samples. Fuel, Vol. 6, Marh 1981, pp. 226-23. 5. J. M. Dereppe, C. Moreauz, and H. Castex. Analysis of Asphaltenes by Carbon and Proton Nulear Magneti Resonane Spetrosopy. Fuel, Vol. 57, July 1978, pp. 435-441. 6. Y. A. Al-Farkh, N. R. El-Rayyes, F. H. Al-Hajjar, and K. A. Al Zaid. Studies on Kuwait Crudes. Strutural Analysis of Asphalts and Their Frations by Nulear Magneti Resonane. Applied Spetrosopy, Vol. 33, No. 4, 1979, pp. 46-49. 7. B. Brule. Charaterization of Bituminous Compounds by Gel Permeation Chromatography (GPC). Journal of Liquid Chromatography, Vol. 2, No. 2, 1979, pp. 165-192. 8. C. Suh, B. Brue!, and C. Baluja-Santos. Charaterization of a Road Asphalt by Chromatographi Tehniques. Journal of Liquid Chromatography, Vol. 2, No. 3, 1979, pp. 437-453. Marathon's objetive in onduting this study was to further its understanding of asphalt properties and manufature. Any use or reliane on data ontained herein shall be at the reader's own risk. Marathon makes no representation, extends no warranties of any kind, express or implied, and assumes no responsibilities whatsoever with respet to the appliation and use of the information ontained herein. Publiation of this paper sponsored by Committee on Charateristis of Bituminous Materials.