Identification of Dicarboxylic Anhydrides in Oxidized Asphalts J. C. Petersen, F. A. Barbour, and S. M. Dorrence Laramie Energy Research Center, Bureau of Mines, U S . Department of the Interior, Laramie, Wyo. 82070
Dicarboxylic anhydrides were identified as a significant product formed on Oxidation In asphalts. Their infrared absorption in the carbonyl region was a broad, overlapping doublet band with major and minor peak maxima at about 1725 and 1765 cm-‘, respectively. The anhydrides possessed considerable stability and their corresponding sodium salts, formed by hydrolysis with sodium hydroxide, readily reverted to the anhydride on acidification. The chemical reactivity of the asphalt anhydrides most nearly matched that of six-membered-ring aromatic anhydrides. The presence of significant amounts of non-ring intermolecular anhydrides was dlscounted. Esters were not found In detectable amounts in oxidized asphalts; anhydrides in asphalt were probably partly responsible for earlier reports of the presence of esters.
Functional group types absorbing in the carbonyl region of the infrared spectra have long been recognized ( 1 - 4 ) as major products of the oxidation of asphalts. These functional groups are a mixture of several different classes of chemical compounds that are not readily distinguishable by their infrared spectra. Although previous investigators have suggested that anhydrides might be present in oxidized asphalts, efforts to confirm their existence were not successful ( I , 5, 6). The present report identifies dicarboxylic anhydrides as a significant oxidation product in asphalts and suggests reasons why these were not identified by previous investigators. The anhydrides were identified by selective chemical reactivity and infrared spectroscopy. The carbonyl infrared bands of the anhydrides and their derivatives were made readily observable by the elimination of the interference of the strong ketone band through the use of differential spectroscopy. A model compound study was conducted to determine the most probable structural type of anhydrides in asphalts. A quantitative method for the determination of anhydrides in oxidized asphalts is found in another paper ( 7 ) . The conflicting reports ( I , 5, 6 ) of the existence of esters in oxidized asphalts are discussed, and evidence is presented that suggests why anhydrides were mistakenly identified as esters.
(THF) was from Eastman Chemical Co. and was protected from moisture by storing over type 4A molecular sieve from J. T. Baker and Company. Hexamethyldisilazane (HMDS) and trimethylchlorosilane (TMCS) were “specially purified grade” from Pierce Chemical Company. Procedure. Complete procedures for reaction of the samples with sodium hydroxide, potas,sium bicarbonate, or a mixture of hexamethyldisilazane (HMDS8)and trimethylchlorosilane (TMCS) are reported elsewhere ( 7 ) . Briefly, dicarboxylic anhydrides and carboxylic acids were converted to their sodium salts by reaction of a 0.25-gram sample of asphalt or asphalt fraction dissolved in 10 ml of benzene with 0.4 ml of 0.5M aqueous sodium hydroxide; the reaction occurs while slowly evaporating (in about 15 minutes) the benzene and water from t,he sample on a hotplate. Carboxylic acids were selectively neutralized by a similar treatment of a duplicate sample with potassium bicarbonate. Silyl esters of the acids or silyl esters of the sodium salts of the anhydrides and acids were prepared by treatment of a 0.15-gram sample in 3 ml of T H F with a mixture of 0.02 ml of HMDS and 0.01 I$ of TMCS for one-half hour a t 40 “C. Spe&al analyses of the silylated samples were made on the reaction mixture without prior separation or solvent removal. The phenylimide of the dicarboxylic anhydrides in oxidized Wilmington saturate fraction was formed by refluxing 0.125 gram of the fraction with freshly distilled aniline for 23 hours under a nitrogen atmosphere. The unreacted aniline was evaporated from the residue a t reflux temperature using an inert gas purge. Residual amounts of aniline were removed by adding and evaporating two 4-ml portions of benzene from the residue. Infrared Spectra. Infrared spectra were obtained using 0.1-cm sealed cells and absorbance-scale chart paper on a Perkin-Elmer Model 621 infrared spectrophotometer. Matched cells were used for differential spectra ( 7 ) ,and a variable cell was used for solvent compensation. All spe‘ctra except those of the model anhydrides were obtained on 0.25-gram samples dissolved in 5.00 ml THF, using solvent compensation. Spectral data for model anhydrides and their corresponding carboxylic acids and silyl esters were obtained on. dilute solutions of unknown concentrations.
RESULTS AND DISCUSSION
Evidence f o r Dicarboxylic Anhydrides i n Asphalt. The reactions that were used in this study to identify and characterize the dicarboxylic anhydrides in oxidized asphalits are illustrated in Figure 1, using l&naphthalic anhydride as a model. These reactions and their significance when carried out on asphalt samples are discussed in detail in the sections that follow. The reactions were followed by o‘bserving changes in the infrared absorption spectra in the carbonyl region from 1500 to 1900 cm-l. An oxidized saturate fraction from a Wilmington (CaliEXPERIMENTAL fornia) asphalt was extensively used for illustrative purposes because it contained a relatively large amount of the Materials. Asphalt 30 was supplied by the Federal Highway Administration (FHWA) and is identified by the number used in proposed dicarboxylic anhydrides relative to other functhe publication ( 8 ) which describes its initial physical properties. tional types absorbing in the carbonyl region and because The Wilmington saturate fraction had been separated from a deasof the virtual absence of any interfering bands prior to oxiphaltened sample of Wilmington (Calif.) asphalt (9-13) by elution dation. In most whole asphalts, the anhydride bands are alwith hexane from a column of basic alumina (11). Asphalt 30 had most completely masked by the intense carbonyl absorpbeen recovered by benzene extraction from an 11-year-old pavetion a t about 1700 cm-’ and appear only as a broadening of ment. The Wilmington saturate fraction had been air oxidized for 24 hours a t 130 “C in a GLC column ( 1 4 ) on a solid support of Fluthe left wing of this band, although the bands are discernoropak 80. Benzo[ghi]perylene-l,2-dicarboxylicanhydride was obible in differential spectra ( 7 ) . The results obtained on the tained from K and K Laboratories, and benzo[ghz]perylene-7,8-di- saturates fraction, however, are qualitatively similar to carboxylic anhydride was prepared in our laboratory by the oxidathose obtained on all other fractions from the asphalt and tive degradation of coronene according to the procedure of Ott and on whole asphalts from the FHWA ( 7 , 8 ) . That the absorpZinke (15 1. The remaining four model dicarboxylic anhydrides tion bands studied in the oxidized saturate fraction result were obtained from Aldrich Chemical Co. Benzene was analyzed reagent from J. T. Baker Chemical Company. Tetrahydrofuran from oxidation of the fraction is evidenced by comparing ANALYTICAL CHEMISTRY, VOL. 47, NO. 1, JANUARY 1975
107
1900
1800
I700
1600
? ¶ k ? k & & d O 0
WAVENUHBER. tm-l
Flgure 3. Differential infrared spectra showing effect of silylation of oxidized Wilmington saturates fraction (A) NaOH-treated vs. NaOH-treated, silylated: (B) silylated vs. unsilylated
Figure 1. Typical reactivity of dicarboxylic anhydrides (1&naphthalic anhydride used for illustration only)
\ I
13bO
'
Id00
'
I&
'
'
I$,
f-.
'
' I q - W Q G WPYENuHBER, cm-1
Figure 2. Infrared spectra showing effects of sodium hydroxide treatment and silylation of oxidized Wilmington saturates fraction (A) unoxidized; (B) oxidized: (C)NaOH-treated, oxidized: (D) differential spectrum of C vs. 6; (E) differential spectrum of NaOH-treated, oxidized vs. silylated, oxidized
spectra A and B, Figure 2 , obtained on the unoxidized and oxidized fractions, respectively. The effects of the treatment of the oxidized saturate fraction with sodium hydroxide are shown by comparing infrared spectra B and C, Figure 2 , which were obtained on samples before and after treatment, respectively. Treatment with sodium hydroxide caused a decrease in the absorption of the fraction in the 1700-1800 cm-' region and a corresponding increase in absorption in the 1550--1650 cm-' region. The decrease at 1700-1800 and the increase at 1550-1650 cm-l are attributed to a shift in the C=O stretching frequency resulting from hydrolysis of the dicarboxylic anhydrides and neutralization of the carboxylic acids to form carboxylate ions as illustrated by the reaction equation Figure 1. The changes in the spectra are best illustrated by differential spectrum D, which is the difference between spectra B and C; spectrum D was produced with a sodium hydroxide-treated sample in the sample beam and an untreated sample in the reference beam of the spectrophotometer. The infrared absorption band attributed to anhydrides in asphalts is the broad doublet band above the null line with a major peak a t about 1725 cm-l and a minor peak or shoulder a t about 1765 cm-l. The dual nature of the carbonyl absorption band of dicarboxylic anhydrides is illustrated by spectral data in Table I. The broad band below the null line in spectrum D and centering at 1580 cm-' results from absorption of the C=O of the carboxylate ion which was derived from the anhydrides and carboxylic acids. Only a small component of the 1725 cm-1 peak in the anhydride doublet band shown in Figure 2 is believed to result from absorption by free carboxylic acids. The free 108
acids are differentiated from the anhydrides by silylation with MHDS and TMCS and by reaction with potassium bicarbonate; the acids react with these reagents and the anhydrides do not (cf. Figure 1). Thus the change from spectrum D to spectrum E in Figure 2 results from silylation of the small amount of acids in the fraction, shifting their carbonyl absorption from that of the free acids a t about 1730 cm-I to that of the corresponding silyl esters a t about 1715 cm-l. This results in an apparent shift of the 1725 cm-' anhydride peak to a slightly lower frequency. The anhydride peak at 1765 cm-l in spectrum E remains viturally unaffected. The above interpretation of spectra D and E is supported by spectra in Figure 3. Differential spectrum A shows the difference between a sodium hydroxide-treated and a sodium hydroxide-treated, silylated sample. Treatment with sodium hydroxide converted the anhydrides to their sodium carboxylates, and subsequent silylation converted the sodium salts of both the anhydrides and acids to silyl esters thus giving rise to the large silyl ester peak at about 1715 cm-I. If the major functional type giving rise to the bands in spectrum E, Figure 2, were free carboxylic acids, or if the anhydrides were directly silylatable, then spectrum E, Figure 2, should be identical to spectrum A, Figure 3. The small amount of free acids in the sample, which silylate and shift the 1725 cm-l band as previously mentioned, are evidenced by the small free acid band a t 1730 cm-' and the small silyl ester band at 1715 cm-l in the silylated us. unsilylated spectrum (spectrum B, Figure 3). The absence of a negative peak at 1765 cm-l in spectrum B, Figure 3,provides additional evidence that the functionality responsible for the 1765 cm-' band is virtually unaffected by silylation. That both the 1725 cm-l and 1765 cm-' bands produced in asphalt on oxidation are associated with the same functional group is evidenced by the data in Table 11; these data show changes in band areas of the differential spectra during stepwise partial hydrolysis of separate, but identical, asphalt samples. The ratios of the band areas are essentially the same after each partial hydrolysis even though differing amounts of sodium hydroxide were used. The larger ratios at the lower sodium hydroxide levels are caused by the small amount of free acids present which are preferentially neutralized and are measured in the 1725 c n - l region. These acids account for about six of the arbitrary band area units. The amounts of sodium hydroxide used in these experiments do not equate stoichiometrically with the amounts of carboxylic acids and dicarboxylic anhydrides. The relative amounts of the excess sodium hydroxide which remain unreactive and/or which react with other components in the asphalt are not known. Another evidence for anhydrides in asphalt is the appar-
ANALYTICAL CHEMISTRY, VOL. 47, NO. 1, JANUARY 1975
Table I. Summary of Spectral Data and Relative Reactivity of Dicarboxylic Anhydrides and Their Derivatives Frequency of carbonyl band, cm-1, in THF solvent
Anhydride Compound
Relative reactivity Free acid
ring size
Anhydride
Re-forms
(silyl ester)Q
Silylates
on acidification
3, 3,4,4-Benzophenonetetra-
5
1885, 1857
1732 (1693)
partial
partial (20%)
carboxylic anhydride Benzo[glzi]perylene- 1, 2dicarboxylic anhydride
5
1768 (shoulder on low frequency side)
1725 (1705)
most
1,8-Naphthalic anhydride 3,4,9,10-Perylenetetracarboxylic anhydride" Benzo[glii]perylene- 7,8 dicarboxylic anhydride Diphenic anhydride
6 6
172Zd (1702)
1730 (1716)
Yes Yes
7
1780, 1775, 1744, 1770,
partial (hydrolyzed easily in THF) no no
1712 (1692)
almost none
no
7
1785.5, 1 7 5 1
small amount
no
Asphalt anhydrides
?
1765, 1725
no
ves
1744.5 1756 1730 1735
1731.5, 1715 (1710, 1693) 1725-1730 (1710-1715)
a Prepared by silylation of sodium salt of corresponding acid. * Under conditions indicated in Experimental. Neutralization of sodium salts with slight excess of dilute HC1 followed by removal of water by azeotroping with benzene. d Readily converts to anhydride on heating during evaporating of solvent. e Sample did not dissolve in THF. Spectrum was obtained on a dispersion of the solids in THF.
Table 11. Changes in the Anhydride Carbonyl Bands during Stepwise Hydrolysis of Anhydrides in 0.25 g of Oxidized Asphalt 30 Area under absorption hands, arbitrary units
MI of 0.l.N' SaOH
0.1 0.2 0.3 0.4 0.5 0.6 0.8
Table 111. Carbonyl Absorption Frequencies of Selected Dicarboxylic Anhydrides and Their Phenylimide Derivatives
Ratio of
Frequenc), cm
-1
band areas,
1765crn-1
172511765 cm-'
4
4 4.5 3.3 2.8 2.7 2.6 2.7
1725 cm-'
11 13 14
14
Anhydride type
1,8-Naphthalic anhydride Wilmington s a t u r a t e s fraction oxidized on quartzite
ent complete reversibility of the sodium hydroxide reaction by acidification as indicated in Figure 1. When the hydrolyzed saturate fraction in benzene was acidified with stoichiometric amounts of hydrochloric acid followed by evaporation of the solvent, the infrared spectrum of the hydrolyzed-reacidified sample was identical with that of the untreated sample. Although free carboxylic acids would show a similar reversible reaction on acidification, their presence in significantly large amounts was ruled out by the silylation reaction. Hydrolysis of esters would have an effect similar to the hydrolysis of anhydrides on the infrared spectrum; however, esters would not be expected to re-form on acidification under the mild conditions used. Consideration will be given later to the possible presence of esters in oxidized asphalts. To further substantiate that anhydrides were involved in the reversible reaction cited above, we attempted to observe the spectra of the free acids during stepwise hydrolysis in T H F of the corresponding silyl esters, prepared from sodium salt of the anhydrides, prior to ring closure to form the anhydrides. The silyl esters were extremely sensitive to the incremental additions of water and were hydrolyzed almost instantaneously a t room temperature. Although some free acids corresponding to the anhydrides were observed, ring closure t o form the anhydrides was apparently highly
Anhydride
Phen) limide
1782,1747 (sharp)
1712,1670-1680
1765,1725 (broad)
(latter a doublet) 1690,1630- 1650 (latter a doublet)
WAVERUMBER. cm.1
Figure 4. Infrared spectra showing phenylimide formation in oxidized Wilmington saturates fraction (A) untreated; (6) after reaction with aniline; (C)differential spectrum of B ws. A
favored. Except in this controlled experiment, the free acids corresponding' to the anhydrides have not been observed in the many asphalts studied to date. The final evidence for the existence of anhydrides in oxidized asphalts is the formation of the phenylimide by reaction with aniline. The reaction is illustrated in Figure 1. Changes in the infrared spectra of the oxidized saturate fraction on imide formation are shown in Figure 4. Spectra
ANALYTICAL CHEMISTRY, VOL. 47, NO. 1, JANUARY 1975
109
Table IV. Changesa in Intensities of Selected Infrared Absorption Bands during Reaction of Aniline with the Dicarboxylic Anhydride in Oxidized Wilmington Saturates Fraction Area
Area
b
under
0
carbonyl
5-mern be r s d ring
anhydride phen) limlde
carbonil band
Loss of (1810-1700
bands
h r at l i O ° C
0 6 23
A
B C
i a r b o n \ l , ’ > trar) units)
0
0
54
49 74
81
...
Peak helght
(1630-1700 of N-phenyl cm‘l,
R e a c t i o n rin’e, a n h ) d r > d e cm-*, arbi- arbltrar)
Run ho.
units)
0
63 10 7 0.59
aromatic
hand
a t 1597 c m - l (arbitrary units)
0 16.5 26
Ratio, B/C 0.66 0.63 aTaken from a differential spectrum similar to spectrum C, Figure 5. b Represents anhydride which has disappeared. A and B were obtained before and after imide formation, respectively, and spectrum C is the differential spectrum. Infrared absorption frequencies of the imides obtained for the saturate fraction and also for pure l&naphthalic anhydride are reported in Table I11 together with the frequencies of the original anhydrides. Although the absorption frequencies for l,%naphthalic anhydride and it,s derivative are higher than those for the anhydrides in the asphalt fraction, the spectra are similar in that both samples showed two major imide bands, the latter being a doublet. The anhydride and imide bands in the asphalt fraction are much broader than those of the pure compounds, as would be expected, because the asphalt anhydrides are undoubtedly a complex mixture of structural types. An additional band also appeared a t 1597 cm-l (spectrum C) which is attributed to absorption by the N-phenyl aromatic ring of the phenylimide. Evidence that the bands attributed to the phenylimide of the asphalt fraction result from a derivative of the anhydride is presented in Table IV. Data in the table compare the increase in the anhydride carbonyl band of the differential spectrum ( c f . Figure 4), which indicates a loss of anhydride, with the increase in the phenylimide carbonyl and N-phenyl aromatic bands; comparisons were made a t 54% and 81% completeness of reaction. The ratios of band intensities, B/C, a t these two stages of the reaction, as seen in the table, are in good agreement considering the expected error in estimating band intensities. Possible Types of Dicarboxylic Anhydrides in Asphalts. Because dicarboxylic anhydrides are produced in greater amounts during the oxidation of asphalts than are free carboxylic acids, oxidation of ring-structure molecules to form the anhydrides is indicated. Random oxidation of noncyclic side chains to form carboxylic acids would produce free carboxylic acid groups having little tendency to form anhydrides. Carboxylic acid groups attached to nonring systems, where molecular configuration is less restricted, would be expected to form anhydrides that are less stable than the asphalt anhydrides and that are not easily reformed by neutralization of the acid salts. Several structural variations of aromatic anhydrides will be considered on the assumption that aromatic dicarboxylic anhydrides are the most probable type formed in asphalts. Aromatic dicarboxylic anhydrides can be classified as those having five-, six-, and seven-membered anhydride rings as illustrated in Figure 5. The relative reactivity of two representative anhydrides of each ring size was exam110
rn
\
under
6-membered ring
7-membered ring
Figure 5. Anhydride types considered as possible types formed in asphalt on oxidation
ined using experimental procedures similar to those used in the study of the anhydrides in asphalt. Results of these studies are reported in Table I. As seen in the table, the chemical reactivity of the sixmember-ring anhydrides most nearly matched that of the anhydrides produced in asphalt. Molecular models of the six-membered-ring anhydride showed almost no strain, indicating a stable ring structure. The six-membered-ring anhydrides did not silylate under the usual silylation conditions, and their sodium salts readily reverted to the anhydride on neutralization with hydrochloric acid and the usual solvent evaporation. The five- and seven-membered-ring anhydrides appear less favored than the six-membered ring anhydride as the anhydride type in asphalts. Unlike the anhydrides in asphalt, the five-membered-ring anhydrides were partially silylated as a result of ring opening. This is not surprising because of the apparent strain in the five-membered anhydride ring, as indicated by molecular models. The hydrolyzed five-membered-ring anhydrides studied did partially re-form, however, after acidification and solvent removal, as previously described for the asphalt fraction. One of the five-membered-ring anhydrides(benzo[ghi]perylene-1,2dicarboxylic anhydride) was readily hydrolyzed to the free acid a t room temperature in T H F solvent containing trace amounts of water, in contrast to the anhydrides in asphalt which re-form from hydrolysis of their silyl esters under similar conditions. Although somewhat resistant to silylation, the seven-membered-ring anhydrides, after hydrolysis with sodium hydroxide, did not easily revert to the anhydride on acidification. This property, characteristic of the anhydrides in asphalts, sets the seven-membered-ring anhydrides apart from the anhydrides in asphalts. The Ester Controversy and Its Relationship to Anhydrides. As previously indicated, the existence of esters in oxidized asphalts has been disputed ( I , 5, 6). When our investigations of oxidized asphalts showed that most of the sodium hydroxide-reactive material absorbing in the carbonyl region was not free carboxylic acids ( 7 ) ,the possibility that esters might account for this material was considered. However, a frequency of 1765 cm-I for one of the bands affected by the sodium hydroxide treatment seemed too high for esters. Also, the observation that the carbonyl spectrum of the sodium hydroxide-treated sample returned to its original condition on simple acidification with hydrochloric acid and solvent evapcration was not consistent with the ester hypothesis because hydrolyzed esters normally do not re-form under such mild conditions. However, an experiment was performed to explore the unlikely possibility of esters reforming on acidification. A model ester, phenyl stearate, was added to an oxidized asphalt and the mixture was treated with sodium hydroxide in the usual manner; this caused complete hydrolysis of the phenyl stearate. Excess 1-naphthol was then added to compensate for possible loss of the more volatile phenol from the phenyl stearate, the mixture was neutralized with hydrochloric
ANALYTICAL C H E M I S T R Y , VOL. 47, NO. 1 , JANUARY 1975
acid, and the solvent was evaporated as usual. This restored the asphalt carbonyl bands to their original condition, but the stearic acid did not re-esterify. Thus, phenyl stearate did not behave in a manner similar to the sodium hydroxide-reactive components in asphalt. Although certain sterically favored lactones might re-form on acidification, these compounds do not form a doublet absorption band in the carbonyl region and are unlikely candidates as significant oxidation products in asphalt. Consideration should be given to the reports by Knotnerus ( I , 5 ) that esters were the major sodium hydroxide-reactive components absorbing in the carbonyl region. He considered, but ruled out, anhydrides as major contributors to this region because he did not observe a significant band. The anhydride doublet band, which is much broader and more overlapping in asphalts than in pure compounds, is difficult to observe because it appears as a weak shoulder on the strong ketone band. In most asphalts the slight inflection caused by the anhydride band is hardly visible, and reaction with sodium hydroxide produces only a shift in high frequency wing of the carbonyl band. Without the use of differential spectra, the anhydride band is understandably missed. The spectra in the Knotnerus paper ( 1 ) show the shift in the high-frequency wing of the carbonyl band produced by the reaction of sodium hydroxide with the anhydrides. However, the broadness of the carbonyl peak indicates that the resolution of the spectrometer was insufficient to detect the inflection on the band wing caused by anhydrides. Although Knotnerus states that the strong carbonyl absorption at 5.8 p (about 1720 cm-1 and presumed by him to be mainly esters) largely disappears on reaction with sodium hydroxide and a new band representative of the C=O vibration of carboxylic acid salts is formed, examination of his spectra shows that after sodium hydroxide treatment most of the 5.8-p band still remains. This is consistent with our work which indicates that the major oxidation products absorbing in the carbonyl region are ketones (7, 1 6 ) . We believe that the new carboxylate band a t 6.3 p (about 1580 cm-l) in Knotnerus’ spectrum can be accounted for largely by the shift or loss in the high-frequency wing of the carbonyl band resulting from anhydride hydrolysis. We believe that what was interpreted as a loss of the 5.8-p carbonyl band was primarily a loss of the low-frequency wing of the band resulting from a broad base-line depression ( 7 ) in the 6.3-/Lregion caused by treatment with sodium hydroxide.
Campbell and Wright (6) disagreed with the conclusion of Knotnerus that esters were major contributors to the carbonyl region of oxidized asphalts, largely on the basis of spectral data obtained by the addition of model esters to oxidized asphalts. Their work showed no apparent reinforcement of observable bands in the asphalts. They eliminated the possibility of the presence of significant amounts of anhydrides by similar experiments with model anhydride additions. We believe that Campbell and Wright were also unable to observe existing anhydride bands in oxidized asphalts because the broad overlapping bands were not distinguishable on the wing of the intense carbonyl band. These bands are less distinct in neat films as used by Campbell and Wright than in T H F solutions used in our work.
ACKNOWLEDGMENT The personal interest and assistance given by Woodrow
J. Halstead, Edward Oglio, and others of the Office of Research, Materials Division, FHWA, are greatly appreciated.
LITERATURE CITED J. Knotnerus, J. lnst. Petrol., 42, 355, (1956). B. D. Beitchman, J. Res. Nat. Bur. Std., Sect. A, 63, 189 (1959). J. R. Wright and P. G. Campbell, J. Appl. Cbem., 12, 256 (1962). C. D. Smith, C. C. Schnetz, and R . S. Hodgson, lnd. Eng. Cbem., Prod. Res. Develop., 5, 153 (1966). J. Knotnerus, ErdoelKoble, Erdgas, Petrochem., 23, 341 (1970). P. G. Campbell and J. R. Wright, J. Res. Nat. Bur. Std., Sect. C, 68, 115 119641. J. C . Petersen, Anal. Cbem., 47, 112 (1975). J. Y. Welborn and W. J. Halstead. Public Roads, 30, 197 (1959); Also Proc. Ass. Asphalt Paving Techno/., 20, 242 (1959). J. C. Petersen, Fuel, 46, 295 (1967). J. C. Petersen, R. V. Barbour, S.M.Dorrence, F. A . Barbour, and R. V. Helm, Anal. Cbem., 43, 1491 (1971). R . V. Barbour and J. C. Petersen, Anal. Chem., 46, 273 (1974). T. C. Davis, J. C. Petersen, and W. E. Haines, Anal. Cbem., 38, 241 (1966). J. W. Rarnsey, F. R. McDonald, and J. C . Petersen, lnd. Eng. Cbem., Prod. Res. Develop., 6, 231 (1967). T. C. Davis and J. C. Petersen, Anal. Cbem., 38, 1938 (1966). R. Ott and A. Zinke, Monatsh., 84, 1132 (1953). S. M. Dorrence, F. A. Barbour, and J. C. Petersen, Anal. Cbem., 46, 2242 (1974).
RECEIVEDfor review March 5 , 1974. Accepted September 6, 1974. The author gratefully acknowledges partial financial support of this work by the Federal Highway Administration in an interagency cooperative program with the Bureau of Mines. Mention of specific brand names or models of equipment is made for information only and does not imply endorsement by the Bureau of Mines.
ANALYTICAL CHEMISTRY, VOL. 47, NO. 1, JANUARY 1975
111