2207
Anal. Chem. 1985, 57,2207-2211
Rapid Isolation of Carboxylic Acids from Petroleum Using High-Performance Liquid Chromatography J o h n B. Green,* B r i a n K. Stierwalt, Jane S. Thomson, and Cheryl A. Treese National Institute for Petroleum and Energy Research, P.O. Box 2128, Bartlesuille, Oklahoma 74005
A normai-phase HPLC method is described for rapid (1 h) separation of carboxylic acids as a class from crude oils and petroleum distillates. Yields of carboxylic acids from several heavy California crude oils as well as mldboillng petroleum distillates generated during the API-60 project are reported. Also, four tricyclic terpenoid acids with m / e values of 320, 334, 376, and 376 have been identified and quantified In carboxylic acid concentrates from several 011s. The general variablilty In carboxylic acid species present In different petroleums is discussed along with the suitability of the proposed method for studies requiring analysis of large numbers of oils.
Table I. Liquid Chromatographic Conditions for Preparative Separation of Carboxylic Acids 30 cm X 2.5 cm (id.) 316 stainless steel Adsorbosil 10 (Alltech Associates, Deerfield, IL) N(average plates/meter) 20000 0.5 cm/min chart speed 35.0 "C temperature 5-10 g of oil in 50-70 mL of amount injected chloroform-hexane (1:9) ca. 80 mL sample loop size Spectra Physics (San Jose, CA) 8000 instrument HPLC A: n-hexane gradient B: chloroform C: methanol with 0.015% tetramethylammonium hydroxide columna packing
Carboxylic acids in petroleum are of interest as a source of biological marker compounds, natural surfactants for enhanced oil recovery methods, and information on geochemical parameters such as maturation or microbiologicaldegradation. Early work concentrated on identification of individual low molecular weight paraffinic acids for purposes of commercial scale isolation and purification ( I ) , while more recent work has been slanted heavily toward geochemical interests (2). Regardless of the focus of these investigations, they have one common feature-a rather lengthy and laborious procedure for isolation and purification of the carboxylic acid concentrate. To the authors' knowledge, the most rapid and complete acid isolation method available requires 8 h for extraction plus additional time for esterification and purification (3, 4 ) . Clearly, practical use of carboxylic acids in geochemical or other types of investigations requiring analysis of large numbers of petroleum samples depends on development of more rapid methods for their isolation from the bulk hydrocarbon matrix. This paper presents a normal-phase HPLC method for isolating carboxylic acids as a class from whole crude oils and distillates in approximately 1h. Selective retention of carboxylic acids on silica is achieved through use of a mobile phase spiked with a strong base, tetramethylammonium hydroxide (5, 6). Sufficient analytical data on resulting carboxylic acid concentrates are provided to demonstrate the viability of the HPLC method and its potential applications, but no systematic attempt was made to generally characterize carboxylic acids in petroleum.
ditions were as follows: initial temperature, 50 "C; 6 OC/min to 350 "C final (10 min hold); injector, 350 "C; detector, 360 "C; flow, 1 mL/min He; 1OO:l split injection; detector makeup gas flow, 29 mL/min. Similar GC conditions were also used during analysis of methyl esters with a Kratos (Ramsey, NJ) MS-80 GC/MS system. This system was comprised of a Carlo Erba Model 4662 temperature programmed GC, jet separator, E1 source, MS-80 magnetic scan mass spectrometer, and Data General Nova 4-based DS-55 data system. Infrared spectra of dilute (10 mg/mL) carbon tetrachloride solutions were recorded with a Perkin-Elmer (Norwalk, CT) Model 283 spectrophotometer using matched 0.5 mm path length NaCl cells.
EXPERIMENTAL SECTION Table I lists equipment and experimental conditions used for preparative scale separation of carboxylic acids. Whole oils were filtered through a sodium sulfate bed and finally a 0.45-pm Teflon membrane for removal of particles and brine prior to dissolution and injection into the HPLC. Because a mass balance for each separation was desired, both fraction 1 (oil minus carboxylic acids) and fraction 2 (carboxylicacids) were collected, stripped of solvent on a rotary evaporator, and weighed. Residual tetramethylammonium hydroxide in the fractions was removed by extraction with aqueous 1 M HCl prior to obtaining final weights. Aliquots of the acid concentrates were esterified with N,Ndimethylformamide dimethylacetalafter the procedure of Thenot et al. (7)GC-FID (flame ionization detector) traces of methyl esters were obtained with a HewlettPackard (Avondale,PA) Model 5880 GC using a J and W (RanchoCordova, CA) 0.25 mm x 30 m, 0.25 pm film thickness DB-5 fused silica column. Instrumental con-
RESULTS AND DISCUSSION HPLC. Figure 1 contrasts HPLC traces of a synthetic blend of acidic and neutral compounds (A), a petroleum distillate containing a relatively high concentration of carboxylic acids (B), and a typical whole crude (C). As expected, the separation between carboxylic acids and other classes is generally less distinct in whole crudes than in distillates because of the wider range of compounds in whole crudes. Also, as seen in Figure lA, the most likely contaminants for carboxylic acid concentrates obtained by this method are weaker acids such as phenols, amides, and pyrrolic benzologues. Aliphatic carboxylic acids elute before aromatic acids, and substituted aromatic acids have intermediate behavior. For example, the vast majority of acids in either petroleum sample in Figure 1elute before the standard aromatic acids in Figure
time, min
%A
%B
0
90 90 90
19 20 30 40 55
30 30 30 30 30
58 68
90 90
50
53
%C
flow, mL/min
10
0
10 10 60 60
0 0 10
10 10
40 40 60 10 10
10 30 30 10 0 0
28 28 28 28 28 28 28 28
"New silica columns required preconditioning with 2 L of methanolic tetramethylammonium hydroxide (0.1%) as well as two to four blank runs through the solvent program prior to use.
0003-2700/85/0357-2207$01.50/00 1985 Amerlcan Chemical Society
2208
ANALYTICAL CHEMISTRY, VOL. 57, NO. 12, OCTOBER 1985
005r
I1
Monomer
V
L L - - L - - 3000
203500
C
2500 2000 WAVE NUMBER, c m -
'
2 I600
I800
Figure 2. Infrared spectra of Wilmington 370-535 OC carboxylic acid concentrates (not esterified) obtained with (A) normal fractionation time and (B) delayed cut point (see arrow, Figure 1B). Fraction yield was (A) 2.31 wt % and (B) 1.47 wt %. Concentration of acids in spectral solution was 6.6 mg/mL for both A and 8. Note the larger aliphatic C-H absorption in (A) vs. the nearly equal carbonyl absorption in A and 6.
Table 11. Yields of Carboxylic Acids from Whole Crude Oils and Distillates Whole Heavy California Oils sample no., field T ME
wt % carboxylic acids
rn n u l w
Figure 1. Liquid chromatograms of (A) synthetic mixture (2 mL sample injection loop used: ca. 0.1 mg of each compound injected), (8) WiImington, CA, 370-535 OC petroleum distlllate, and (C) Lompoc 4886 whole crude (B and C; 80 mL sample loop; 10 g of oil dissolved in 65 mL of 9: 1 hexane:chloroform). Key for peaks numbered in (A): (1) dibenzofuran, (2) benzophenone, (3) 2,4,5-trimethylphenoI, (4) 13Hdibenzo[a ,/]carbazole, (5) oxlndoie, (6) 0-cresoi, (7) 3,4-dimethylphenol, (8) phenol, (9) 2-naphthol, (10) 2,2diphenylpropanolc acid, (11) p-toluic acid, (12) 1-fluorenecarboxylic acid, (la), 1-naphthoic acid. Arrow in (B) denotes cut point for aromatic acid fraction illustrated in Figure 28.
1A. Also, this conclusion was verified by means of dual runs of Wilmington 370-535 "C distillate incorporating different cut points. Run 1 utilized the normal cut point whereas run 2 was deliberately cut late (see arrow, Figure 1B). The resulting fraction yields were 2.31 and 1.47 wt %, respectively. Figure 2 shows infrared spectra of the two acid concentrates. The C-H stretching region in the spectrum of the acids obtained using the later cut point shows a decreased aliphatic response per unit weight, which implies higher aromaticity which in turn correlates with the increased UV response of late eluting acids shown in Figure 1B. This conclusion has special significance to the experimenter interested in studying biological marker carboxylic acids, which are nearly all aliphatic, because particular care must be exercised in establishing an appropriate fractionation time which minimizes contamination of carboxylic acids with other types yet partitions the early eluting saturated acids into the appropriate fraction. Table I1 shows yield data for carboxylic acids in whole heavy California crude oils as well as selected petroleum distillates. The whole crudes were chosen to represent widely varying geological classes and the distillates were fractionated in order to compare with existing data on carboxylic acid content (8). The distillates themselves and compositional data on each were products of the API-60 project. In that study, carboxylic acid content was determined by infrared measurements and/or other methods on whole acid concentrates isolated with nonaqueous ion exchange chromatography (9). Results on carboxylic acid content in Wilmington and Cerro Negro distillates were also determined in this laboratory from sub-
7072, South Elwood 7076, South Elwood 7077, South Elwood 7081, South Elwood 4886, Lompoc 4881, Casmalia 4882, Santa Maria Valley 4922, Santa Maria Valley 4924, Santa Maria Valley 7044, Santa Maria Valley 7045, Santa Maria Valley 7049, Santa Maria Valley 4918, Cat Canyon 4926, Cat Canyon 4927, Cat Canyon
1.2
1.0 0.96 2.4 0.79 2.6 1.0 1.1
2.3 1.6 1.9 0.88 1.0 1.4 1.2
Distillates origin
boiling range, O C
Cerro Negro, Venezuela Wilmington, CA Gach Saran, Iran Prudhoe Bay, Alaska S. Swan Hill, Canada Recluse, Wyoming
425-550 370-535 370-535 370-535 370-535 370-535
wt % carboxylic acids 4.0"
-b
2.3 0.16 0.13 0.13 0.13
3.5 0.10 0.18 0.11 0.19
4.05c 2.45 -
-
"This work, direct isolation from distillate. Data from McKay et al. (8). Ion exchange resins were used for isolation of acids; see text. CThiswork, ion exchange resins were used to isolate acids, followed by subfractionation into compound classes on silica. fractionation of whole acid concentrates isolated using nonaqueous ion exchange (10). They also are shown in Table I1 for comparison. Generally, yields of carboxylic acids from the different methods agree well-considering the precision and accuracy typical of separation and gravimetric procedures. GC and GC/MS Analysis of Acid Esters. GC-FID traces were obtained on all acid fractions listed in Table 11. Figure 3 illustrates three of the most widely divergent traces obtained. Figure 3A, typical of immature oils like Wilmington and Cerro Negro, shows a broad hump with relatively few well resolved peaks rising above the continuum. Subsequent GC/MS analysis revealed a fragmentation pattern in these well-defined peaks characteristic of tricyclic terpenoid acid esters (2,11,
ANALYTICAL CHEMISTRY, VOL. 57, NO. 12, OCTOBER 1985
2209
Table 111. Concentration (wt % ) of Tricyclic Terpenoid Acid Methyl Esters in Whole Oil and Distillate Acid Concentrates" peak no.h
1
2
334
3 376
4 376
nd nd nd nd nd 0.01 nd nd nd nd nd nd nd nd nd
0.071 nd nd nd nd 0.24 nd nd nd 0.037 0.15 0.019 nd nd nd
0.025 nd nd nd nd 0.072 nd 0.011 nd 0.17 0.21 0.006 nd nd nd
0.022 nd nd nd nd 0.074 nd 0.010 nd 0.18 0.22 0.006 nd nd nd
0.02 0.03 0.05 0.04 0.07 nd nd nd nd
0.38 0.42 0.63
0.21 0.20 0.21 nd 0.22 nd nd nd nd
0.20 0.21 0.22 nd 0.22 nd nd nd nd
320
mle
heavy California oils 7072, South Elwood 7076, South Elwood 7077, South Elwood 7081, South Elwood 7886, Lompoc 4881, Casmalia 4882, Santa Maria Valley 4922, Santa Maria Valley 4924, Santa Maria Valley 7044, Santa Maria Valley 7045, Santa Maria Valley 7049, Santa Maria Valley 4918, Cat Canyon 4926, Cat Canyon 4927, Cat Canyon distillates (boiling range, "C) Cerro Negro, Venezuela' (425-550) Cerro Negro, Venezuelae (425-550) Wilmington, CAc (370-535) Wilmington, CAd (370-535) Wilmington, CA' (370-353) Gach Saran, Iran (370-535) Rudhoe Bay, Alaska (370-535) South Swan Hills, Canada (370-535) Recluse, WY (370-535)
0.10
0.63 nd 0.091 nd nd
211
2/(3 + 4) 1.5
24
1.6 0
0.11 0.35 1.6
19 14 13 2.5
0.9 1.0 1.5
9
1.4 1.4
aMultiply by data in Table I1 to get concentration in whole oil or distillate. nd = not detected. bPeak numbers refer to Figure 3A. 2 / 1 and 2/(3 + 4) are ratios of peak areas for the respective peak numbers. cWilmington carboxylic acid concentrate obtained with usual procedure. Wilmington carboxylic acid concentrate obtained using delayed cut point; see Figures 1B and 2B. e Wilmington carboxylic acid concentrate obtained by using ion exchange resins followed by subfractionation to obtain carboxylic acid subfraction. 12). The most prominent ion produced from fragmentation of this family has mle 191. The probable structures for peak
~
_
_
_
Table I V . Analysis of Cat Canyon 4927 Methyl Esters I. Identified and Quantified
peak no." 1 2
rn/e = 191
1 ( m l e 320), peak 2 ( m / e )334), and peaks 3 and 4 ( m l e 376) are shown below. These compounds were both common
n
fYY w m/e = 3 2 0
Pf' C02 Me
3 4
compoundb
wt % in acid concentrate
ClS fatty acid methyl ester C16fatty acid methyl ester C17 fatty acid methyl ester C19fatty acid methyl ester
0.035 0.017 0.11 0.09
11. Identified Only, 10.01 wt % Concentration A. straight chain fatty acid methyl esters: C1o, C1l, Clz, C13 B. branched chain (isoprenoid?) fatty acid methyl esters: Clo, C119 ClZ, c13 C. monounsaturated fatty acid methyl esters: C9, Clo, C14 111. Probable Contaminants dioctyl phthalate (from solvents)
n=l
rn/e=334
n =4
rn/e=376
enough and geochemically significant to warrant their quantitation in all carboxylic acid concentrates. Thus, Table I11 lists results for the four compounds as determined by GC-FID (octadecane internal standard) with periodic GC/MS verification of peak identities. Several trends were evident from these data: (1) the excellent agreement in levels of terpenoid acids found in Wilmington and Cerro Negro carboxylic acid concentrates isolated by HPLC vs. those obtained via ion exchange chromatography followed by subfractionation, and (2) the much lower levels of terpenoid acids, especially the complete absence of the m l e 376 doublet, in the Wilmington 370-535 OC carboxylic acid concentrate obtained with the delayed cut point (see Figures 1B and 2B). This latter finding further substantiates the point made earlier concerning the relatively early elution time of saturated carboxylic acids vs. aromatic acids. One additional point revealed by Table III is that some acid concentrates contained no detectable amounts of m l e 191
Osee Figure 3B. bCarbon numbers given include methyl ester carbon. precursors. GC-FID chromatograms of these samples generally looked entirely different from those rich in terpenoid species, as may be seen by comparing Figures 3A and 3B. Prominent peaks in these acid concentrates were usually fatty acid methyl esters. Table IV gives a semiquantitative accounting of major components in the Cat Canyon 4927 esterified acid concentrate. Carboxylic acid esters from most of the MI-60 distillates (except Wilmington, of course) as well as several of the heavy California oils exhibited GC-FID patterns similar to those in Figure 3B. The large predominance of odd-numbered methyl esters (even-numbered acids) indicated in Table IV was typical and has been observed elsewhere (2, 13). Finally, some samples contained mixtures of terpenoid, fatty acid, and other acid subclasses. Figure 3C shows an example of that type of GC-FID trace and Table V lists tentatively identified major peaks indicated on Figure 3C. As shown in
2210
ANALYTICAL CHEMISTRY, VOL. 57, NO. 12, OCTOBER 1985 Table V. Analysis of Santa M a r i a Valley 7044 M e t h y l Esters
I. Identified and Quantified
peak mol no.‘ wt 1
2 3 4 5
6 7
8 9 10
wt % in acid
compoundb
164 Cl0-monoaromaticacid methyl ester 198 C,,-monocyclic saturated acid methyl ester 240 CI6-monocyclicsaturated acid methyl ester 270 C17-fattyacid methyl ester 298 CI9-fattyacid methyl ester 334 tricyclic terpenoid acid methyl esterC 376 tricyclic terpenoid acid methyl ester 376 tricyclic terpenoid acid methyl ester 404 tricyclic terpenoid acid methyl ester 404 tricyclic terpenoid acid methyl ester
concentrate 0.23 0.21 0.08 0.05 0.05
0.037 0.17 0.17
0.029 0.029
11. Identified only, 10.05 wt % concentration 262 dimethyl cyclopentylphthalate 276 dimethyl cyclohexylphthalate 111. Probable Contaminants dioctyl phthalate (from solvents) ‘See Figure 3C. *Carbonnumbers and molecular weight include methyl ester carbon(s). CSeetext.
TIME
m
nul*(
-
Figure 3. GC-FID traces of (A) Wiimington 370-535 OC (normal cut point), (B) Cat Canyon 4927, and (C) Santa Maria Valley 7044 esterified acid concentrates. See text for GC conditions. Peaks eluting before 8 min are artifacts from the esterification reaction. Numbered peaks in A-C are identified in Tables 111-V, respectively. is = internal standard (octadecane). DOTH = dioctyl phthalate (solvent impurity).
the table, the Santa Maria Valley 7044 acid esters contained at least two previously unobserved m / e 404 precursors to the terpenoid m l e 191 fragment ion discussed earlier. Also listed are several abundant acids with cycloalkane and/or aromatic nuclei. This type of acid concentrate appears to be intermediate between the fatty acid dominated concentrates from mature or possibly biodegraded oils and the “broad-hump” pattern with relatively few resolved components characteristic of geologically immature oils. Assessment of Method and Quality of Fractions. A statistical evaluation of the method’s precision and accuracy is not presented because of the lack of reference fuels and other standards as well as the multiple end uses of the separation. For example, repetitive runs of the same sample under identical HPLC conditions resulted in carboxylic acid yields with a f 5 % relative average deviation. However, this deviation results mainly from factors such as losses during sample handling and solvent removal. Many potential users of this method would be interested in the concentrations of a few key compounds, not the overall yield of carboxylic acids. For those users, the &5% relative average deviation from gravimetric determinations would not be relevant, since they would not be isolating and weighing acid concentrates. Rather, their overall precision would be limited by esterification reproducibility and GC or GC/MS errors. The authors attempted to address these points by separating the API-60 distillates, for which literature data on carboxylic acid content were available, and also by determining
carboxylic acid concentrations with nonaqueous ion exchange chromatography followed by subfractionation of the material retained by the ion exchange resin. The agreement between the carboxylic acid yield data from the two methods was encouraging but not definitive. Similarly, results for tricyclic terpenoid acid methyl ester concentrations (Table 111) from concentrates obtained by HPLC and ion exchange agreed very well, but equal agreement for other components cannot be assumed. Thus, the available data reflect favorably on the method‘s precision and accuracy, but rigorous analysis of these parameters is impossible a t the present time. Quality or purity of the carboxylic acid concentrates is also difficult to determine. For example, analysis of the major components, such as in Tables 111-V, accounted for less than 1% of any given concentrate. On the other hand, analytical methods such as IR, NMR, or elemental analysis, which yield average data on the whole sample, are usually not sensitive to minor amounts of noncarboxylic acid impurities. For example, the elemental composition of the Wilmington 370-535 “C acid concentrate with normal cut point shown in Figures 1B and 2A was 76.3% C, 10.6% H, 0.10% N, 1.6% S, and 11.4% 0 (by difference). Thus, the sulfur and nitrogen could be present either from impurities or from thiophenic carboxylic acids/thiocarboxylic acidslpyridinic carboxylic acidsletc. The infrared spectrum (Figure 2) shows no evidence of other compound types, but IR has poor sensitivity for many potential contaminants. Concentrates from oils containing very low levels (