ANALYTICAL CHEMISTRY, VOL. 50, NO.
CONCLUSION T h e short I3C relaxation times of crude oil residues investigated enable the use of pulse methods for obtaining I3C NMR spectra applicable to quantitative analyses without an addition of a relaxation reagent. Some differences between 'H and 13C NMR values follow however, from a comparison A more general of the relative amounts of C,, and C,,. evaluation of both methods will require characterization of many samples of various origin.
(3) (4) (5) (6) (7) (8) (9) 110) (1 I j (12)
6,MAY 1978
775
Ed., Academic Press, New York, N.Y., 1973, p 429. H. C. Dorn and D. L. Wooton, Anal. Chem., 48, 2147 (1976). L. Thiault and A. Mersseman, Org. Magn. Reson., 8, 28 (1976). J. N. Schoolery, Progr. Nucl. Magn. Reson., 11, No. 2 (1977). M. Saas and D. Ziessow, J . Magn. Reson.. 25, 263 (1977). T. F. Yen, L. J. Boucher, J. P. Dickie, E. C. Tynan, and G. E. Vaughan, J . Inst. Pet., 55, 87 (1969). R. H. Filby, Am. Chem. SOC., Div. Pet. Chem., Prepr.. 18, 630 (1973). R. E. Williams, Spectrochim. Acta, 14, 24 (1959). J. G. Soeiaht. Fuel.. 49. 76 (1970). T. F. Yen, G. Erdman, and A.'J. Saraceno, Anal. Chem., 34, 695 (1962). H. L. Retcofsky, F. K. Schweighardt, and M. Hough, Anal. Chem., 49, 589 (1977).
LITERATURE CITED (1) R. Freeman, H,O.w. Hill, and R. Kaptein, J . Magn, Reson,, 7 , 327 (1972). (2) W. De W. Horrocks, "NMR of Paramagnetic Molecules", G. N. LaMar,
RECEIVED for review September 27, 1977. Accepted January
11,1978.
Prediction of Capacity Factors for Aqueous Organic Solutes Adsorbed on a Porous Acrylic Resin E. M. Thurman," I?. L. Malcolm, and G. R. Aiken Water Resources Division,
U.S.
Geological Survey, Denver Federal Center, Denver, Colorado 80225
The capacity factors of 20 aromatic, aliphatic, and alicyclic organic solutes with carboxyl, hydroxyl, amine, and methyl functional groups were determined on Amberlite XAD-8, a porous acrylic resin. The logarithm of the capacity factor, k', correlated inversely with the logarithm of the aqueous molar solubility with significance of less than 0.001. The log k'-log solubility relationship may be used to predict the capacity of any organic solute for XAD-8 using only the solubility of the solute. The prediction is useful as a guide for determining the proper ratio of sample to column size in the preconcentration of organic solutes from water. The inverse relationship of solubility and capacity is due to the unfavorable entropy of solution of organic solutes which affects both solubility and sorption.
Amberlite XAD-8 is a macroreticular, nonionic, acrylic ester polymer ( I ) . Because of ester cross-linkage, it has a more polar structure than Amberlite XAD-2 (styrene-divinylbenzene polymer) which has been used extensively to preconcentrate trace organic solutes from natural waters (2-10). In spite of t h e widespread use of XAD resins, there has been no comprehensive effort to model which solutes will adsorb and what their respective capacity factors will be. Rather, recovery data for classes of compounds have been compiled ( 1 , 4 , 7). Some work has been done to elucidate the general adsorptive properties of the resins ( I , 4,111; van der Waals' forces, dipole interactions, and hydrogen bonding are thought to be important ( I ) . This paper examines the effect of functional group and carbon skeleton on t h e capacity factor of an organic solute on XAD-8 resin in order to model solute retention. Our work shows that solubility can be used to predict capacity to a first approximation using a log h'-log solubility plot. T h e importance of this finding is t h a t solubility, a simple physical parameter, can be used to match column and sample size for t h e most effective concentration of the solute of interest. This paper not subject to U.S. Copyright.
Our interest in XAD-8 resin arises from our work with natural water samples where we have found t h a t XAD-8 is much more efficient than XAD-2 in the recovery of natural humic substances (12-14). Humic substances account for approximately 25% of the total dissolved organic carbon (DOC) in natural waters.
EXPERIMENTAL Resin. Amberlite XAD-8 was obtained from Rohm and Haas and prepared for column studies by washing the 20/50 mesh beads with equal volumes of 0.1 N NaOH for five successive days to remove monomers and soluble, uncross-linked polymers. The resin was then Soxhlet extracted sequentially with methanol, acetonitrile, and diethyl ether for a 24-h period each and stored in methanol. Reagents. Butanoic, pentanoic, hexanoic, cyclohexylcarboxylic acids; n-butanol, n-pentanol, n-hexanol, cyclohexanol, p methylbenzyl alcohol; cyclohexylamine and hexylamine were obtained from Aldrich. Benzaldehyde, benzyl alcohol, and toluene were obtained from Baker. Benzoic, phthalic, and toluic acids were obtained from Eastman. Aniline, benzene, and phenol were obtained from Mallinckrodt. All reagents were analyzed reagent grade. Reagent water was prepared by passing distilled water through a mixed bed, ultra-pure, ion-exchange resin. The pH 2-reagent water was prepared by lowering the pH of reagent water with concentrated HC1, then passing it through a precolumn of Amberlite XAD-8 resin. Instrumentation. A Varian Techtron UV-VIS scanning spectrophotometer Model 635, a Varian gas chromatograph 2700 Series with a flame ionization detector, and a Beckman 915 carbon analyzer with a Model 215 A infrared detector were used for standard and eluate analysis. Glenco glass columns, 3500 Series, 0.9 X 30 cm, with Teflon fittings and tubing, and a Technicon AutoAnalyzer proportionating pump were used for adsorption chromatography. Procedure. Stock solutions of 200-500 mg/L were prepared by dissolving 100-250 mg of the organic compound in 500 mL of reagent water. The concentration of the stock solution was checked by DOC analysis. Stock solutions were diluted to 1.2 x M or approximately 10 mg/L DOC with pH 2 water for column adsorption experiments (isotherms were approximately linear but more importantly these concentrations are environPublished 1978 by the American Chemical Society
776
ANALYTICAL C H E M I S T R Y , VOL. 50, NO. 6, M A Y 1978
Table I. Capacity Factor on XAD-8 Compounds
’
Aniline Benzaldehyde Benzene Benzoic acid Benzyl alcohol p-Methylbenzyl alcohol Butanol Butanoic acid Cyclohexanol Cyclohexylamine Cyclohexylcarboxylic acid Heptanoic acid Hexanoic acid Hexanol Hexylamine Pentanoic acid Pent anol Phenol Phthalic acid Toluene p-Toluic acid
Table 11. Comparison of Solute k’ Values k’
126 337 452 488 123 317 25 39 75 78 390 960 377 266 256 125 93 245 144 1406 1037
mentally realistic for DOC). All standard solutions were pumped onto the columns as nonionic species. The pH was adjusted to two pH units below pK, or two pH units above pKb for ionic solutes to suppress ionization. The resin was packed into glass columns from a water-methanol slurry. The column was rinsed with approximately 200 column volumes of reagent water or until the DOC was equal to the blank (less than 0.3 mg/L DOC). The column was cleaned with two 25-mL alternate rinses of 0.1 N NaOH and 0.1 iK HC1. The pH of the column was adjusted to that of the standard solution with reagent water. Capacity factor, k ’, was determined by frontal chromatography. The k’determinations were precise within * 7 % . This precision was attained because large amounts of solutes (10-80 mg) were adsorbed and could be measured accurately by DOC analysis. These high loadings correspond to conditions of analysis of natural water samples where organic carbon levels range from 5-20 mg/L. Various organic substances in natural waters which are present at trace levels compete for adsorption sites and are adsorbed together. It is for this reason that test solutes were determined at concentrations of 5-10 mg/L DOC rather than trace levels. Standard solutions were pumped onto the column at a rate of 4 mL/min (10 bed volumes per hour) until effluent and influent concentrations were equal. Then columns were eluted with an appropriate solvent and the eluates quantified by gas chromatography (methanol eluates) or DOC analysis (aqueous eluates). The 50% breakthrough point was used t o check eluate recovery. The capacity factor was calculated as the mass of solute sorbed on the resin divided by mass of solute present in the void volume of the column. I t was found that flow rates exceeding 20 bed volumes per hour caused a decrease in capacity. This is consistent with the work of Simpson ( I ) . This dependence of k ’upon flow rates exceeding 20 bed volumes per hour suggested that the adsorption process at these flow rates may be a nonequilibrium condition and k‘s may in fact be apparent k’s. However. a flow rate of 10 bed volumes per hour or less gives constant values of h’(which appear to be the result of an equilibrium process), is a conservative flow rate for preconcentration purposes, and is realistic for practical considerations when using natural water samples.
RESULTS AND DISCUSSION T h e capacity factors for the 20 solutes studied are listed in Table I. The differences among h ’ measurements indicate that the resin apparently favors aliphatic over aromatic over alicyclic carbon systems. For instance, compare heptanoic acid ( k ’ = 960),benzoic acid (h’ = 488),and cyclohexylcarboxylic acid (h’= 390); then n-hexapol, phenol. and cyclohexanol; and n-hexylamine, aniline, and cyclohexylamine (Table IIA). Likewise, the following functional groups were preferred -CH3 > -C02H > -CHO > -OH 2 NH, (Table IIB). Notice that
Part A. Comparison of carbon skeleton.
CG2 h
C32b
960
488
OH
OH
266
245
75
256
390
I
I
“2
“2
126
I8
Part B. Comparison of Functional Groups
OH
25
0
38
OH
0
90
125
266
37 7
NH
256
0 9Q Q NH
OH
78
75
NH
0H
126
245
CH2GH
123
C HO
337
C02H
48 8
CH3
1406
the resin preferences follow an inverse solubility trend. That is. in general, aliphatic compounds are less soluble than aromatic compounds which are less soluble than alicyclic compounds. Also, the functional group trend follows decreasing solubility. This resin preference can be best understood by comparing the molar aqueous solubility of the solutes. When the logarithm of the aqueous molar solubility for the 20 diverse solutes listed in Table I were plotted vs. log h’, a rather well defined linear relationship was found with a correlation coefficient of 0.9 (significant a t less than 0.001). See Figure 1. The equation is log k ‘ = 1.77 - 0.52 log S.The identical capacity factor of several solutes for XAD-8 emphasizes the importance of solubility. Aniline, benzyl alcohol, and pentanoic acid have the same molar solubility and identical capacity factors. Also this result was noted for benzene and benzoic acid. This relationship of log solubility
ANALYTICAL CHEMISTRY, VOL. 50, NO. 6, M A Y 1978
777
Table 111. Log D-Log Solubility Relationship on XAD-2 Resina Compound
I C L -
1
-2 I
- 3
3
lag Solubility
Figure 1. Log k’-log solubility plot
t o log h’ emphasized t h a t although absolute retention is determined both by eluent and stationary phase, the interaction between solute and solvent decides selectivity for XAD-8 resin. A similar result was found by Locke (15) on a different reverse phase system. He states that if an adsorbent is homogeneous with only weak interactions with solutes, then the relative retention of functionally similar solutes is determined by their relative solubilities in the mobile phase. Locke (15)found a log h’-log solubility relationship for three of four substituted ureas chromatographed on a C-18 bonded packing with 10% methanol and water as the eluent. He also used the log h’plot to predict the solubility of slightly soluble compounds. Others have noted solubility and capacity relationships. Log h’or R , has been related to log solubility in partition chromatography (16). T h e logarithm of the partition coefficient (octanol/water) has been related to log solubility (17, 18). Finally, Karger and others (29) state that log h’,,, (h’in a mixed water and organic solvent system) is proportional to log water solubility in a reverse phase system. This relationship between capacity and solubility is attributed to the hydrophobic effect as reviewed by Karger and others (19). Simply, nonpolar organic solutes have an unfavorable entropy of solution which is due to ordering of water molecules around the organic solute (20). This negative entropy is a driving force both for aggregation and removal of the nonpolar solutes and is called the hydrophobic effect. T h e log h’-log solubility relationship can he derived as a linear free energy diagram because log h’is proportional to the free energy of adsorption of the solute, and log solubility is proportional to t h e free energy of solvation. Thus it can be seen that the unfavorable entropy of solution of organic solutes is related to both the free energy of adsorption and to the free energy of solvation; this is one explanation of the log h’-log solubility relationship. Because distribution coefficients have been measured on XAD resins by others ( 4 ) ,we checked to see if the inverse solubility trend was present in their work. Twelve distribution coefficients measured by Grieser and Pietrzyk ( 4 ) on XAD-2 in 10% ethanol and water (for which solubility data were available) gave an inverse linear log D-log solubility plot. The compounds are listed in Table 111. The regression coefficient was -0.87 and was significant a t 0.001. Also similar work being completed in our laboratory suggests that other XAD-resins (1,2 , 4 and 7 ) show a similar log h’-log solubility relationship. Finally, it appears t h a t solubility can be used to predict capacity factors for diverse organic solutes of low molecular weight when preconcentrating them on nonionic Amberlite XAD resins. Free Energy of Adsorption of a Methylene Group. For a homologous series of surface active substances, there is a
o-Nitrophenol m-Nitrophenol p-NitrophenolC o-Chlorophenol m-Chlorophenol p-Chlorophenol 2,4-Dichlorophenol 2,4,6-Trichlorophenol Phenol Benzene Methylbenzoate Chlorobenzene
log Db
2.81 2.05 1.80 2.20 2.40 2.29 2.93 3.32 1.63 2.52 3.09 2.92
Log solubilityd
-1.82 -1.01 - 0.94 -0.65 - 0.69 - 0.68 -1.55 -2.39 -0.15 -1.69 -2.94 -2.36
a Data from Grieser and Pietrzyk ( 4 ) . D determined Corrected D f o r partial ioniin 10% ethanol and water. zation. Solubility data (26). Correlation is - 0.87.
constant ratio between molar concentration of two successive homologues which gives rise to an equal lowering of surface tension in aqueous solutions; this is known as Traube’s rule (21). The surface activity increases strongly and regularly as a series is ascended (22). Langmuir (23) attributed this change in surface activity to the decrease in partial molar free energy of the solute in the surface layer. From theoretical considerations he calculated that the free energy of adsorption should increase by 710 cal/mol/carbon atom for a series of aliphatic acids. Butler (24) presented convincing arguments that the free energy of adsorption was equal to the summation of several partial molar free energies. One of which is the soluteesolvent interaction term which is related only to solubility. He showed that the other terms remained constant while ascending the homologous series and that the change in the free energy of adsorption within the series is related only t o the change in solubility. Butler presented aqueous solubility data for a series of ethyl esters of aliphatic acids which showed that the difference in free energy of solvation ascending the series has a mean of 640 cal/mol/carbon atom. This value is in accord with Langmuir’s theoretical estimate of 710 cal/mol/carbon atom. T h e free energy of adsorption of a methylene group on Amberlite XAD-8 was determined for a homologous series of aliphatic alcohols and acids, and for t h e addition of a methylene group to an aromatic acid, aromatic alcohol, and an aromatic hydrocarbon. Because the free energy of adsorption is proportional to log h’, the difference in log h’for two solutes which differ by only a methylene group can he used to compute the free energy of adsorption that is associated with the addition of a methylene group. The mean free energy of adsorption a t 20 “C was 646 cal/mol/carbon atom for the aliphatic system and 550 cal/mol/carbon atom for the aromatic system (Table IV). These results are similar to the previously cited solubility estimate of Butler and the theoretical estimate of Langmuir. Although the free energy difference between compounds is small, h’is altered from 30 to 90 (butanol to pentanol) and 90 to 270 (pentanol to hexanol). During the preconcentration of organic solutes from a 1-L water sample by 6 g of resin, this corresponds to differences in retention of the solute from 10% to 5070 to 9570, respectively. Effect of pH and Ionic Strength on log k’log solubility plot. Many have noted the dramatic drop in h’ upon the ionization of an organic solute ( I , 4 , 25); and in order for preconcentration to occur on the Amberlite XAD resins, ionization of the solute must be suppressed ( 1 , 4 ) . For organic acids the preconcentration must be done a t a p H two units below pK,; and for organic bases a t a pH two units above pKb.
778
ANALYTICAL CHEMISTRY, VOL. 50, NO. 6, MAY 1978
Table IV. Free Energy of Adsorption of a Methylene Group at 20 "C on XAD-8
Compound
log k '
A G of adsorption Alog k' of -CH,
Aliphatic system Butanol Pentanol Hexanol Butanoic acid Pentanoic acid Hexanoic acid Heptanoic acid Mean
1.40 1.97 2.42 1.59 2.10 2.58 2.98
...
...
...
0.57 0.45
7 64 60 3
0.51
684 644 536 646
...
0.48 0.40 0.49
*..
Aromatic system Benzene Toluene p-Methylbenzyl alcohol Benzyl alcohol Benzoic acid Toluic acid Mean
2.66 3.15 2.50 2.09 2.69 3.02
...
...
...
0.49
657
0.41
550
0.33 0.41
442 550
...
...
...
...
For example, the h ' of benzoic acid dropped from 475 to approximately 1 when ionization occurred. In terms of the solute-solvent interaction, this indicates that solubility increased owing to the ion-dipole interaction of the solute and the water molecules (that is, the solubility of sodium benzoate is many times greater than undissociated benzoic acid). Thus a compound must be nonionic to use the log k'-log solubility plot. Ionic strength does increase the k ' of the hydrophobic organic solute; and In h 'has been shown to be related to ionic strength ( 4 , 25). However, over the range of ionic strengths of most natural waters (less than 0.01), the effect on the capacity factor is small and does not significantly affect the log h'-log solubility plot. Use of log &log Solubility Plot. Although the capacity-solubility relationship in reverse phase chromatography is not a new finding, it is a very expedient one for preconcentration purposes on XAD-resins. I t is now possible to predict the capacity factor to a first approximation for any organic solute from natural waters on XAD-8 resin. This allows one to match column and sample size for the most efficient removal of the organic solute of interest. Log solubility can be used to estimate h'and even qualitative data can be of some value. For example, the solubility code used in chemical handbooks can be used to estimate retention. We have found that compounds which are insoluble or slightly soluble are completely retained (6-g column and 1-L sample), those which are soluble to very soluble are partially retained and those which are very soluble to miscible are not retained. We use the term hydrophobic to describe those solutes which are completely retained, and hydrophilic to describe those which are not retained. Quantitative solubilities should be used whenever possible. Solubilities for this study were obtained from the "CRC Handbook of Chemistry and Physics" (26). Other possible sources include references 27-29. T o estimate h'of a particular solute, the molar solubility is substituted into the regression line, log k ' = 1.77 - 0.52 log S. From breakthrough curves of the 20 test solutes, we found t h a t h'can be related empirically to bed volume and sample size (30). T h e equation is:
X = '/sk'Y where X is volume of sample in mL, is a constant, and Y is bed volume of the column in cm3. This empirical equation
simply relates the amount of sample t o resin volume before breakthrough occurs. Using this equation, a solute with a h ' of 300 would be completely retained (no breakthrough) on a 20-mL column (6 g of resin) from a 1-L sample. SYith this equation and the log h'-log solubility plot, resin t o sample ratios can be adjusted for maximum retention of the solute of interest. A corollary is that given the amount of resin and sample used in preconcentration, one can estimate the h'or solubility of solutes completely retained by the resin. This is the practical importance of the log h '-log solubility plot. It was noted t h a t breakthrough for a solute with a k'of 300 occurred earlier than was expected from the relationship of V , = V , (I + h ) where V , is the retention volume of the solute and Vo is the void volume of the column. We attribute this early appearance of the solute to eddy diffusion around the large macroporous resin beads. This results in diffuse breakthrough fronts in the column. The DOC of most natural waters varies from 5 to 20 mg/L (31). Approximately 50% of this DOC can be removed by Amberlite XAD resins (14, 30). These various solutes all compete for adsorption sites on the resin, therefore, h 'values were determined a t loadings of 10 m g / L DOC rather than trace loadings from pure solution. This is an important consideration when preconcentrating organic solutes from water, and points out an area of concern in the previous work of others in the isolation of organic solutes from water. There is a misconception that organic solutes can continue to be concentrated on a few grams of macroreticular resin while many gallons of water are being passed through the column. Solutes such as benzene or benzoic acid will leak through a 6-g column of Amberlite XAD resin (which is larger than commonly used) after applying only a few liters of sample. Even very insoluble compounds such as polynuclear aromatic hydrocarbons have a limited h ' a n d must compete with more numerous natural organic solutes for adsorption sites. Based on work with model solutes on the Amberlite XAD-8 resin, 1 L appears to be an ideal volume of sample to use for preconcentration of polar and nonpolar organic solutes on a 6-g column.
LITERATURE CITED R. Simpson, "The Separation of Organic Chemicals from Water", Rohm and Haas, Philadelphia, Pa., 1972.J. P. Riley and D. Taylor, Anal. Chim Acta, 46, 307 (1969). A. K. Burnham. G. V. Caider. J. S. Fritz. G. A. Junk. H. J. Svec. and R. Wiilis, Anal. Chem., 44, 139 (1972). M. D. Grieser and D. J. Pietrzyk. Anal. Chem., 45, 1348 (1973). J. J. Richard and J. S. Fritz, Talanta. 21, 91 (1974). A. K. Burnham, G. V. Calder, J. S Fritz, G. A. Junk, H. J. Svec, and R. Vick. J . Am. Wafer Works Assoc., 65, 722 (1973). G. A. Junk, J. J. Richard, M. D.Grieser, D Witiak, J. L. Witiak, M. D. Arguello, R. Vick, H. J. Svec, J. S. Fritz, and G. V. Calder, J . Chromatogr., 99, 745 (1974). R. F. C. Mantoura and J. P. Riley, Anal. Chim. Acta, 76, 97 (1975). H. F. Walton, Anal. Chem., 48, 52R (1976). J. A. Leenheer and E. W. D. Huffman, Jr., J . Res. U . S Geol. Surv., 4, 737 (1976). R. L. Gustafson, R. L. Aibright, J. Heisler, J A. Lirio, and 0. T. Reid, Jr., Ind. Eng. Chem., Prod. Res. Dev.. 7 , 107 (1968). J. Leenheer, unpublished work, U S Geoi. Sulvey, Denver, Colo., 1976-77. G. R. Aiken, R. L. Malcolm, and E. M. Thurman, Abstract at the 1977 Soil Science Society of America Meeting. E. M. Thurman, G. R. Aiken, and R. L. Malcolm, Proceedings of 4th Joint Conference on Sensing Environmental Pollutants, 1977. 0. C. Locke, J . Chromatogr. S o , 12, 433 (1974). E. Soczewinski and J. Kuczynski, Sep. Sci., 2 , 133 (1968). C. Hansch, J. E. Quinlan, and G. L. Lawrence, J . Org. Chem , 33, 347 (1968). C.T. Chiou, V. H. Freed, D. W Schrnedding. and R. L. Kohnert, Environ. Sci. Techno/.,11, 475 (1977). 6. L. Karger, J. R. Gant, A. Hartkopf, and P. H. Weiner, J . Chromatogr., 128. 67 119761. C. Tanford, "The Hydrophobic Effect: Formation of Micelles and Biological Membranes", Wiley Interscience, New York, N.Y.. 1973. J. Traube, Ann. Chem., 265, 27 (1891). J. J. Kipling, "Adsorption from Solutions of Non-electrolytes", Academic Press, New York, N.Y., 1965. 1. Langmuir, J . A m . Chem. Soc , 40, 1360 (1919). J. A. V. Butler, R o c . R . SOC. London, Ser. A , 135, 348 (1932). C.Horvath, W. Melander, and I . Molnar. Anal. Chem., 49, 142 (1977).
A N A L Y T I C A L CHEMISTRY, VOL. 50, NO. 6, M A Y 1978 (26) C. D. Hodgman, R. C. Weast, R. S. Shankland, and S.M. Selby, "CRC Handbook of Chemistry and Physics", Chemical Rubber Co., Cleveland, Ohio, 1961. (27) H. Stephen and T. Stephen, "Solubilities of Inorganic and Organic Compounds", Vol. 1, Part 1, Macmillan, New York, N.Y., 1963-64. (28) J. Pollock and R. Stevens, "Dictionary of Organic Compounds", 5 vols, Oxford University Press, New York, N.Y., 1965. (29) F. K . Beilstein. "Handbuch der Organischen Chemie", J. Springer, Berlin, 1918. (30) R. L. Malcolm, E. M. Thurrnan, and G. R. Aiken, Proceedings of the 11th
779
Annual Conference on Trace Substances in Environmental Health, 1977. (31) R. L. Malcolm and W. H. Durum, U . S . Geol. Surv. Water-Supply Pap., 1817-F, 1976.
RECEIVED for review J u n e 6, 1977. Accepted February 14, 1978. T h e use of brand names in this report is for identification purposes only and does not imply endorsement by the U.S. Geological Survey.
Trapping and Determination of Labile Compounds in the Gas Phase of Cigarette Smoke Steven G. Zeldes and Arthur D. Horton" Analytical Chemistry Division, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37830
The gas phase of cigarette smoke was trapped and stored on Tenax-GC for subsequent off-site analyses. Specifically, the highly labile compounds isoprene, acetaldehyde, and acrolein were determined quantitatively in the samples which were thermally desorbed in the injector port of a gas chromatograph onto a cooled gas chromatographic column. Optimum conditions were determined for adsorption and desorption of the gas phase, and the effects of aging on the trapped gases were studied.
It is necessary to chemically characterize the cigarette smoke offered experimental animals in inhalation bioassays to define the extent and quality of the exposure. It is also of interest to determine the chemical nature of smoke-polluted environments to assess the possible impact of smoking on nonsmokers. Tenax adsorption followed by thermal desorption and gas chromatography has been evaluated as a method for characterizing the volatile organic gas phase constituents of smoke. One of the routine analyses of the gas phase of cigarette smoke is t h e determination of isoprene, acetaldehyde. and acrolein ( I ) ; the first because of its close correlation to biological activity of smoke, and the others because of their ciliatoxicity. If these highly labile compounds can he trapped and retained, then the less labile components of interest should most likely be retained also. Breakthrough volumes have been determined ( 3 . 5 , 8 )for a number of compounds, some of which appear in cigarette smoke. Double trapping experiments have shown that our results agree with those authors for compounds of common interest. Tenax-GC has been used with some success to trap labile compounds in automobile exhausts (2, 31, ambient air (3,4 ) , and stack gases ( 5 ) . T h e traps used for these samples differ only in size, each consisting of a Pyrex tube packed with Tenax held in place by glass wool plugs. T h e methodology used a t this laboratory was adapted from that of Zlatkis, Lichtenstein, and Tishbee ( 4 ) who used a Pyrex tube 11 cm long, 10-mm o.d. and 8-mm i.d. packed with 2 mL of 35 to 60 mesh Tenax. Samples were adsorbed through a condenser and desorbed in a modified injector port onto a cold precolumn, then desorbed a second time onto a n open tuhular column. At this laboratory, the traps were desorbed in the modified injector port directly onto a packed column cooled to -70 "C. This paper not subject to U S . Copyright.
EXPERIMENTAL Adsorbent. Tenax-GC (Applied Science Laboratories, Inc., State College. Pa.), a porous polymer, puly-p-2,6-diphenyIphenylene oxide was selected over other common adsorbents (Porapak, Carbosieve, or activated charcoal) for its several advantages. Its high temperature limit of 450 "C ( 6 ) and low retention volumes ( i )allow high-boiling sample components to lie desorbed more rapidly than from other adsorbents. In addition, the effect of water vapor on the efficiency of Tenax (8) is insignificant. Preparation of Traps. Traps consisted of Pyrex glass tubing (9-mm 0.d.. 5-mm i.d.) cut into 5'/,-inch lengths and fire-polished at each end. One end is ground to a taper to form a seal in the s-inch glass wool plug is placed in the tube at one end, the tube filled. while vibrating. with 60/80mesh Tenax then topped with a "!,-inch glass wool plug. Traps are conditioned by heating at 250 "C for 30 min while purging with nitrogen. Conditioned traps are stored in a desiccator. Sampling Procedure. Samples were collected from weight selected (1094 f 20 mg) Kentucky Reference (1R1) cigarettes conditioned at 75 OF and 6 0 7 ~relative humidity using an ORNL Single Port Smoking Machine. See Figure 1. Cigarettes are smoked at a rate of 1 puff per minute I 11 puffs X 35 mI,/put'f) using a small vacuum pump to draw each puff through a 0.38-mL sampling loop. An additional length of tubing is placed before the inlet to the gas sampling valve in order to collect the sample from the middle of the puff. Nitrogen carrier gas flows through '/,-inch Teflon tubing to a solenoid valve which, when activated, directs the flow through the sample loop and, when deactivated, allows the flow to bypass the loop while the puff is drawn. The carrier then travels through connecting tubing to a stainless steel three-way valve where it is either directed to the modified injector port ( 4 ) of the Perkin-Elmer 3920 gas chromatograph or to a sampling port to which a trap is attached. See Figure 1. During a standardizing run (without trapping) (Figure 2 ) , the carrier containing the sample is directed to the injector port of the gas chromatograph. A simulated trap. filled completely with glass wool, is placed in the injector port to reduce its volume. A number of such runs serves to establish the expected level of organic components in an average cigarette. For a run in which the sample will first be trapped on Tenax. then analyzed (Figure 3): a reduced carrier flow (10 mL/min) is used to purge the sample from the loop into the tapered end of the trap. Carrier gas flow was 1 2 mL/min (30 psig). Injector port temperature was 250 "C; and FID temperature 150 "C. Column. GC Column. The column used for the determination of isoprene, acetaldehyde, and acrolein was a modification of one used routinely ( 1 ) for this purpose. The stationary phase 3,3'-(trimethylenedioxy)dipropionitrile was synthesized by equilibrating a 1:2 mixture of acrylonitrile and 1,3-propanediol Published 1978 by the American Chemical Society
