Anal. Chem. 1987, 59, 2583-2586 2583

(3) Lee, M. L.; Vassilaros, D. L.; WhL, C. M.; Novotny, M. Anal. Chem.

(4) Vassilaros. D. L.; Kong, R. C.; Later, D. W.; Lee, M. L. J. Chromafogr.

(5) Cram, S. P.; Brown, A. C. 111; Frebs, E.; Majors, R. E.; Johnson, E. L. Abstract 115, Pittsburgh Conference on Analytlcai Chemistry and Ap- plied Spectroscopy. Cleveland, OH, 1979.

(6) Apffel. J. A., Jr. Ph.D. Dissertation. Virginia Polytechnic Institute and State University, Blacksburg, VA, 1982.

(7) Grob, K., Jr.; Frohlich, D.; Schilling, B.; Neukom, H.; Nageii, P. J. Chromatogr. 1984, 295. 55-61.

(8) Gfob, K., Jr.; Neukom, H.; Etter, R. J. Chromafogr. 1988, 357,

(9) Grob, K., Jr.; Schilling, B. HRC CC, J. High Resolut. Chromafogr. Chromatogr. Commun. 1985, 8 , 726-733.

(10) Grob, K., Jr.; Walder, Ch.; Schilling, B. HRC CC, J. High Resolut. Chromatogr. Chromatogr. Commun. 1986, 9 , 95-110.

(11) Grob, K., Jr.; Stoil, J. HRC CC, J. High Resolut. Chromatogr. Chro- mafogr. Commun. 1986, 9 , 518-523.

(12) Munari, F.: Trlsclani, A.; Mapelii. 0.; Trestlanu, S.; Grob, K., Jr.; Colin, J. M. HRC CC, J . High Resolut. Chromafogr. Chromatogr. Commun.

(13) Cortes, H. J.; Pfelffer, C. D.; Richter, B. E. HRC CC, J. H/gh Resolut. Chromatogr. Chromatogr. Commun. 1985, 8 , 469-474.

1979, 51 I 768-773.

1982, 252, 1-20.

416-422.

1985, 8 , 601-608.

(14) Cortes, H. J.; Richter, B. E.; Pfelffer, C. D.; Jensen, D. E. J. Chroma-

(15) Philip, C. V.; Anthony, R. G. J. Chrometogr. Scl. 1986, 24, 438-443. (16) Grob, K., Jr.; Karrer, G.; Riekkoia, M.-L. J. Chromatogr. 1985, 334,

(17) Wllllams, P. T.; Bartie, K. D.; Andrews, 0. E. SAE Annual Meeting, Detroit, MI, Feb 1987.

(18) Hopper, M. L. J. Chmmafogr. 1984, 302, 205-219. (19) Williams. P. T.; Bartle, K. D., unpublished work. (20) May, W. E.; Wise, S. A. Anal. Chem. 1984, 56, 225-232. (21) Davies. I. L.; Bartie, K. D.; Williams, P. T.; Andrews, G. E., submitted

for pubiicatlon In Anal. Chem. (22) Schwartz, H. E.; Berry, V. V. LCMag. 1985, 3 , 110-124.

fwr. 1985, 349, 55-61.

129-155.

RECEIVED for review April 13,1987. Accepted July 13, 1987. Financial support for this work was obtained from Science and Engineering Research Council Grants GR/C/59208, GR/C/59144, and GR/D/67415. This work was presented at the 38th Pittsburgh Conference and Exposition on Ana- lytical Chemistry, paper 086, Atlantic City, NJ, March 9,1987.

Spectator Ion Indirect Photometric Detection of Aliphatic Anionic Surfactants Separated by Reversed-Phase High-Performance Liquid Chromatography

J. A. Boiani Department of Chemistry, State University College at Geneseo, Geneseo, New York 14454

A simple technique Is described for the detection of linear alkyl sulfate surfactants after separatlon by reversed-phase high-performance liquid chromatography. It makes use of nonretalned lnorganlc absorber Ions and an ultravlolet de- tector. The optlmum conditions for isocratlc and gradlent elution are discussed and typical examples are given. The theoretical efflclency of the method is estlmated. The tech- nlque b found to give llnear responses for samples contalnlng up to 100 pg of surfactant and has detection lknlts of 3 5 pg depending on sample type.

Anionic surfactants including alkyl sulfates are used widely in cleaning agents, cosmetics, and manufacturing processes. Their analysis by reversed-phase high-performance liquid chromatography (HPLC) is significant because it permits the resolution of aqueous samples of homologues and isomers within the same ionic group. Detection of aliphatic alkyl surfactants presents a problem however because their lack of chromophores prevents the direct use of ultraviolet (UV)- visible detection. Several methods have been employed to overcome this problem including postcolumn derivatization followed by absorbance ( I ) or fluorescence (2) detection, conductivity detection (3), and paired ion chromatography (PIC) with counterion (4) or competing ion (5) absorbance detection. All these methods add to the complexity of the analysis by requiring specialized detectors or, in the case of PIC, more complex retention mechanisms and special reagents which can shorten column lifetime.

The procedure described in this paper utilizes the spectator ions necessary to adjust the ionic strength of the eluent for a modified form of vacancy indirect photometric detection.

With this method the readily available UV detector can be used and the retention mechanisms for reversed-phase sep- arations are in force. The method is evaluated by studying the analysis of a range of linear aliphatic alkyl sulfate sur- factants under various chromatographic conditions.

Vacancy indirect photometric detection is a liquid chro- matography detection method for transparent ionic analytes. A constant concentration of an absorbing ion of like charge is maintained in the eluent and the effluent absorbance is monitored continuously. Wherever analyte ions are located in the eluent, there will be a decrease in absorber ion con- centration in order to maintain electroneutrality. These de- creases appear as negative peaks, or dips, in the background absorbance when an analyte ion passes through the detector cell. The areas of these peaks are a function of analyte ion concentration and absorbing ion absorptivity. The technique was first used in ion exchange chromatography and was de- scribed in detail by Small and Miller (6) who named it indirect photometric chromatography (IPC). The technique has been extended to paired ion chromatography as well (7).

In each of the above cases the absorber ion competes with the analyte ion for sites on the stationary phase and there is an equivalent decrease in absorber ion concentration corre- sponding to the analyte ion concentration it displaces from the stationary phase. Such a mechanism is not at work in the present technique because the inorganic ions used for ab- sorbers are not retained by the stationary reversed phases. Nevertheless, Coulombic repulsion between the anions in the mobile phase will cause some displacement of absorber ions in the column regions where the analyte ions are concentrated. The eluent ionic strength necessary for reasonable retentions of the anionic surfactants lowers the displacement efficiency; however it will be shown that the present method gives a

0003-2700/87/0359-2583$01.50/0 0 1987 American Chemical Society

2584 ANALYTICAL CHEMISTRY, VOL. 59, NO. 21, NOVEMBER 1, 1987

reasonable detection limit and linear response range.

EXPERIMENTAL SECTION Reagents. HPLC grade water, methanol, and acetonitrile

obtained from J. T. Baker Chemical Co., Phillipsburg, NJ, or Fisher Scientific, Chemical Manufacturing Division, Fairlawn, NJ, were used for the mobile phases and for sample solvents. Octyl, decyl, dodecyl, and tetradecyl sodium sulfates, 95-98% pure were obtained from the Eastman Kodak Co., Rochester, NY. All other compounds used were ACS Reagent grade.

Apparatus. The liquid chromatograph used was a ternary gradient IBM LC/9533 with an IBM LC/9523 variable-wavelength UV detector with 1-cm detector cell path length and a Rheodyne 7125 injector. The data were collected on a Linear 2020 10-mV strip chart recorder. Most of the analyses were carried out on a 4.6 X 250 mm octadecyl reversed-phase column supplied by IBM Instruments, Inc., Danbury, CT. The remainder were done on a 4.6 X 250 mm octyl column also supplied by IBM Instruments. Mobile phase absorbance spectra were measured with a Hew- lett-Packard 8451A diode array spectrometer.

Chromatographic Conditions. The mobile phases used ranged from 6040 to 3070 acetonitrile/water and 9010 to 4060 methanol/water. All mobile phases contained 0.01 M dibasic sodium hydrogen phosphate to control pH and either 0.01 M sodium nitrate or sodium iodide to provide the absorbing spectator ion and contribute to the control of ionic strength. Mobile phase pHs ranged from 6.8 to 4.8, the highest being for 9O:lO metha- nol/water and the lowest for 30:70 acetonitrile/water. Actual percent compositions used for each separation were obtained from mixtures of the extreme solution compositions employing the solvent programmer available on the liquid chromatograph.

The absorbances of the extreme solutions were matched to the nearest A0.005 AU by the addition of appropriate amounts of sodium nitrate or sodium iodide. The wavelengths monitored were 242 nm for all nitrate-containing mobile phases, 252 nm for methanol/water iodide mobile phases, and 260 nm for aceto- nitrile/water iodide systems. Typical absorbances were 0.90 AU. This background absorbance was cancelled out by filling the detector reference cell with the most polar mobile phase in each set and zeroing out any residual signal electronically. This pro- cedure allowed for the use of more sensitive detector settings. The negative peaks related to the analytes were measured on the 0.005 or 0.01 AUFS ranges.

A 20-pL sample loop was used for all injections as well as a 1.0 mL/min flow rate. The analyses were done at ambient tem- perature. Sample concentrations ranged from 1.0 to 5.0 mg/mL (lo00 to 5000 ppm). Samples were dissolved in the most aqueous mobile phase used for each eluent system. For separation test runs two surfactant mixtures were used. One contained 1.0 mg/mL of each compound and the other contained 0.0063 M of each compound which in units of mg/mL ranged from 1.4 for octyl sodium sulfate to 2.0 for tetradecyl sodium sulfate.

RESULTS AND DISCUSSION Mobile Phase Absorber Ions. Nitrate and iodide were

the inorganic ions chosen as absorbers because of their stability under analysis conditions and because of their absorbance in the UV spectral region below 250 nm for nitrate and 270 nm for iodide. The concentration chosen was dictated primarily by the need to maintain eluent ionic strength high enough to allow reasonable retention times for the organic anions in question. The studies of the reversed-phase separation of such anionic surfactants done by Smedes e t al. (2) show that the capacity factors of these compounds are strongly dependent on eluent ionic strength until a plateau is reached a t strengths produced by ionic concentrations of the order of M. Small and Miller (6) have shown for IPC that detection sensitivity improves with lower absorber concentrations, so concentra- tions much above the threshold of the plateau region should be avoided. As a compromise between these two criteria, a total ionic concentration of 0.04 M was chosen made up of equal contributions of the 0.01 M sodium dibasic hydrogen phosphate and 0.01 M of the sodium salt of the absorber ion.

The choice of analytical wavelength is also a compromise.

0.0001 AU

DODCCYL

Tlme (minutes)

Flgure 1. Isocratic separation of octyi, decyi, and dodecyl sodium sulfate: column, IBM octadecyi; flow rate, 1.0 mL/min; detector 0.005 AUFS at 242 nm; injection, 20 pL of 1.0 mg/mL solution; mobile phase, acetonitrile/water (4258) containing 0.0 1 M sodium nitrate and 0.01 M sodium dibasic hydrogen phosphate.

Higher sensitivity can be expected for wavelengths with larger absorptivities; but the background absorbance needs to be cancelled out by filling the detector reference cell with the mobile phase, and absorbances larger than 1.00 AU in the reference cell cause a decrease in detector sensitivity. Ab- sorbances in the region of 0.90 AU appeared to be the opti- mum choice for the present system.

The absorbance spectrum of each inorganic ion is dependent to varying degrees on the identity and concentration of organic modifiers. The nitrate absorption maximum shifts to shorter wavelength with increased polarity of the mobile phase. This is the expected behavior for such n - P* transitions as found in nitrate (8). The result is more noticeable at the analytical wavelength, which occurs on the long wavelength slope of the absorption band. The effect is greater for the acetonitrile solution where the absorbance at 242 nm decreases by 8% when going from 30:70 to 6040 acetonitrile/water. For methanol there is a 4% decrease from 40:60 to 9O:lO metha- nol/water. The effect is much more drastic for iodide solu- tions. For acetonitrile, there is a 52% absorbance decrease between the 30:70 and 60:40 acetonitrile/water extremes a t 260 nm. For methanol there is 29% decrease between the 4060 and 90:lO extremes at 252 nm. Such behavior has been reported previously (9) although the reason for it is not as straightforward as for the nitrate absorption.

The concentration dependence of absorbance does not affect isocratic analyses but presents problems for gradient elution. As a convenience when isocratic eluent percentages were changed, the two extreme mobile phase concentrations of each solvent were absorbance matched by adding more absorber compound to one of them. By itself this procedure still did not allow for gradient elution because it was impractical to obtain solvents whose absorbances agree more closely than the 0.005 AU absorbance range that was often used to obtain reasonable negative peak areas. A remedy to this problem is discussed later.

Chromatographic Results. A typical isocratic chroma- togram, using an acetonitrile/water mobile phase, is shown in Figure 1. As expected, acetonitrile was found to be the stronger organic modifier and lower percentages of it than of methanol are needed for obtaining similar retention times. For instance, the same sample as shown in Figure 1, separated by use of a 67:33 methanol/water mobile phase, gave retention times of 4.0, 7.7, and 18.6 min for octyl, decyl, and dodecyl sodium sulfates, respectively.

The retention times of the four test compounds varied too widely on the octadecyl column to obtain a reasonable isocratic separation of more than three of the series at any mobile phase composition. This represents a five-carbon number spread. I t was possible to obtain an efficient separation of a four- component mixture, representing a seven-carbon number

ANALYTICAL CHEMISTRY, VOL. 59, NO. 21, NOVEMBER 1, 1987 2585

I I

/ I 11 j j DODECYL I 0.001 AU

I , DECYL

OCTYL I 1 I I I I I I I

‘ I I I‘

- I

- 0 . 7 2 ~ 0 ~

spread, on the octyl column. Figure 2 is an example of such a separation.

Surfactants with carbon chains shorter than eight were not studied; however, it would seem feasible that they could be separated by the present method with a higher percentage of water in the mobile phase and the separation could be made efficient enough to resolve surfactants differing in only one carbon atom. Also it would be possible to obtain shorter retention times for the longer carbon chain surfactants by using a shorter carbon chain stationary phase, such as methyl or butyl, although this was not tested in the present study. Shortening retention times by increasing the percent organic modifier much above the values used in this work is not reasonable because of the solubility limits of the dissolved Salts.

To perform gradient elutions, the absorbance variation with eluent composition must be cancelled out. For this purpose a third solvent, water, was introduced as a diluent. Ternary gradients were designed where the percentage of water varied at a rate that would dilute the two absorber-containing sol- vents sufficiently to maintain a constant overall background absorbance. In order to avoid major changes in mobile phase composition, the absorbances of the other two solvents were matched as closely as possible before use. In practice, a reasonable binary gradient was designed with reference to predetermined retention times of the analytes and run without sample injection on a higher absorbance range. From the rate of change of absorbance with elution time and a knowledge of the molar absorptivity of the absorber ion, a ternary gra- dient was designed to compensate for this rate of change. The process proved most successful for the methmol/water nitrate mobile phase and Figure 3 shows the result of such a gradient elution on the octadecyl column.

The choice of absorber ion did not appear to have a major impact on the optimum chromatographic conditions, with the exception of dictating analytical wavelength. Iodide-con- taining mobile phases tended to give slightly longer retention times but this may be an ionic strength effect due to the larger amount of iodide that had to be added to match the absor- bances of the two extreme solvent concentrations. Peak areas were found to be comparable for the two ions. For gradient elution however iodide-containing mobile phases did not work as well as those containing nitrate. This is due to the much larger absorbance variation with organic modifier concen- tration found with the former ion.

The mobile phases used had no detectable effect on the efficiency of the column. Once it was washed with a metha- nol/water solvent, it gave a chromatogram for the uracil, phenol, acetophenone, nitrobenzene, methyl benzoate, toluene reversed-phase test mixture similar to that obtained 6 months

0.001 AU T L 1 / 1 1

0 2 4 6 8 10 12 14 16 18

Time (minutes)

Flgure 3. Gradient separation of octyl, decyl, dodecyl, and tetradecyl sodium sulfates: column, IBM octadecyl; flow rate, 1.0 mL/min; de- tector 0.01 AUFS at 242 nm; injection, 20 pL of 0.0063 M solution; solvent A, methanol/water (30:70) containing 0.01 M NaN03 and 0.01 M NaH,PO,; solvent B, methanol/water (9O:lO) containing 0.01 M NaNO, and 0.01 M NaH,PO,; solvent C, water; gradient 100% A at 0.0 rnin to 96.7% B + 3.3% C at 12.0 min.

Arbitrary Units

Analyte In dotmctor cell

at mmximum absorbance

exiting

Analyte distrlbutlon

column

2586 Anal. Chem. 1987, 59, 2586-2593

transparent dihydrogen phosphate anion are displaced, the concentration of displaced absorber ion would be 0.00025 M. Typically 0.01 M solutions of these ions have absorbances of 0.90 AU at the analytical wavelength. Therefore the maximum absorbance change expected would be 0.023 AU. Absorbances observed for such samples were 0.0020-0.0025 AU. Therefore the present detection method is approximately 10% as effi- cient in displacing absorber ions as methods using competing absorber ions with ion exchange separations.

An experimental confirmation of the difference in sensitivity can be found in a recent paper by Larson and Pfeiffer ( IO) on the determination of quaternary ammonium compounds by indirect photometric chromatography, using a 0.01 M so- lution of competing ion (benzyltrimethylammonium chloride) with comparable molar absorptivity to the ions in the present study. The absorbance changes as estimated from their published chromatograms were of the order of 0.02 AU and their reported detection limit of 0.6-0.8 pg was somewhat lower than that reported below.

Sensitivity and Linear Limit. Each of the four surfac- tants gave a linear response up to the maximum amount of 100 wg. The detection limit (SIN = 2.5) was found to be 3-5 pg. This value is somewhat higher than the 0.1-1 pg range stated for the conductivity detector (3) and both are higher than those cited for the postcolumn derivatization techniques where absorption detection ( I ) gives values of 0.01-0.02 pg and fluorescence detection gives values of 0.001-0.005 pg. Nevertheless, the simplicity of the present method compen- sates for this difference in detection limits and the limits are sufficiently low to permit analysis of commercial surfactant mixtures and surfactants in commercial products such as shampoos where sample dilution would be necessary in any case. The addition of a preconcentration column can serve to lower the detection limit for the present method when necessary to analyze for surfactants in the trace amount region.

Also the upper limit to linearity of the present method is 25 times as large as that reported for the postcolumn fluorescence method, the only one reporting such a value.

CONCLUSION Spectator ion indirect photometric detection is a viable

alternative to other techniques for anionic surfactant analyses by reversed-phase HPLC. Its simplicity of operation and linear response at higher concentrations compensates for its somewhat higher detection limit and it can be adapted for use with gradient elution.

ACKNOWLEDGMENT The author wishes to thank three students who helped in

the preliminary research for this project; C. J. Banach, P. Michelsen, and D. Fritzinger.

Registry No. NaN03, 7631-99-4; NaI, 7681-82-5; octyl sodium sulfate, 142-31-4; decyl sodium sulfate, 142-87-0; tetradecyl sodium sulfate, 1191-50-0; dodecyl sodium sulfate, 151-21-3.

LITERATURE CITED Wall, D. Paper presented at the Northeast Regional Chromatography Discussion Group, Rochester, NY, May 20, 1985. Smedes, F.; Kraak, J. C.: Werk, C. F.; hengoewie, H.; Brinkman, U. A. Th; Frei, R. W. J . Chromatogr. 1882, 247, 123-132. Manufacturer's literature, Wescan Instruments, Inc., Santa Clara, CA, 1986. Denkert, M.; Hackzell, L.; Schill, G.; Sjogren, E. J . Chromatogr. 1081,

Bidllngmeyer, B. A.; Warren, V. F., Jr. Anal. Chem. 1082, 54, 235 1-2358. Small, H.: Miller. T. E. Anal. Chem. 1082, 54, 463-469. Barber, W. E.; Carr, P. W. J . Chromatogr. 1983, 260, 89-96. Skoog, D. A. In PTimipIes of Instrumental Anatysls , 3rd ed.; Saunders College Publishing: New York. 1985: p 186. Jsrgensen, Chr. K. In Halogen Chemkhy; Gutman, V., Ed.; Academic: London, 1967: Vol. 1, pp 274-284. Larson, J. R.; Pfeiffer, C. D. Anal. Chem. 1083, 55, 393-396.

278, 31-43.

RECEIVED for review March 23,1987. Accepted June 23,1987.

Correlation of Liquid Chromatographic Retention Times with Molecular Structures of Coal Liquid Components

Francis P. Burke* and Richard A. Winschel Consolidation Coal Company, Research & Development, Library, Pennsylvania 15129

The retention t lnw of the components of a dktillable coal llquid (bp ca. 200-500 "C), separated by reversed-phase liquid chromatography (RPLC) wlng a C18 column, can be esthted based on a mu#lple llnear regredon correlation with 11 slmple molecular structure descriptors. A stepwise methanoi/water gradlent was employed. The coal liquid components Include alkylated and unakylated aromatics, hydroaromatka and other n- heterocycles, and phenols. A correlatkm coeltld.nt of 0.989 was obtabred. The predlcted and observed retention tknes of the 153 com- ponents and groups d Isomen, d" by an average of 2.6 min in a 138-min RPLC separatbn. Improved accuracy (f2.0 mln average) was obtained for a set of 37 model coal liquid compounds.

In the various processes for the direct liquefaction of coal, a process-generated recycle solvent is employed to dissolve

the coal. This recycle solvent consists, a t least in part, of a distillate oil boiling between about 200 and 500 "C. Recycle solvents are highly complex mixtures of alkylated and unal- kylated aromatics, hydroaromatics and other naphthenoaro- matics, heterocyclics, phenols, naphthenes, and paraffins. Since the composition of this recycle solvent can have a sig- nificant effect on the performance of the coal liquefaction process ( I ) , it is desirable to have a detailed understanding of its composition. Gas chromatography/mass spectrometry (GC/MS) is a useful means to accomplish this for distillates; however, these recycle oils are typically much too complex to allow complete separation of components even with very high resolution capillary columns. Fractionation of the distillate recycle oil by liquid chromatography (LC) prior to GC/MS analysis allows a much more thorough characterization of the recycle oil becam each fraction is less complex than the whole sample. In addition, it is possible to exploit the different selectivities of the LC and GC columns to make the GC separation more complete. For example, prefractionation with

0003-2700/87/0359-2588$01.50/0 0 1987 American Chemical Society