Separation of polyaromatic hydrocarbons by liquid-solid

Mar 1, 1973 - Igor F. Perepichka, Lyudmila G. Kuz'mina, Dmitrii F. Perepichka, Martin R. Bryce, Leonid M. Goldenberg, Anatolii F. Popov, and Judith A...
0 downloads 0 Views 501KB Size
governed mainly by diffusion from the gas to the adsorbed phase. Calibration curves were prepared for some acidic compounds eluted on FT-G 0.3% FFAP. These curves showed that linear elution is obtained even if very low amounts of acids are injected. The minimum amount of acid detectable depends in practice on the sensitivity of the experimental apparatus. The potential of GLS columns for the analysis of acidic compounds was tested by chromatographing the lower members of the carboxylic acids. In Figure 5 , a chromatogram corresponding to the elution of a water solution of C?-Cs acids on FT-G 4- 0.3% F F A P is given. As can be seen, an initial peak before acetic acid appears when injecting water solutions. As already reported (9), this is due to a slight decomposition of the F F A P by water. In our case, this effect was attenuated by the use of a low-coated GLS column and injecting small amounts of water mixtures. The absence of peak tailing made it possible to eluate 1 p1 of a water mixture, containing about 40 ppm of each acid. Moreover, a near base-line separation in a short elution time was obtained. This was so because of the high separation factors coupled with only a small loss of performance at high linear gas velocities. However, the major novelty to be noted in this chromatogram is the complete separation of the 2-methylbutyric, 3methylbutyric acid pair. The difficulty in separating this pair by gas-solution chromatography has already been discussed. The capacity of the carbon surface t o separate according to geometric differences in the molecules has also been taken advantage of in separating isomers of some aromatic acids. The chromatogram obtained is shown in Figure 6. Also in this case, FT-G coated with 0.3% F F A P was used. Higher surface coverages give shorter elution times, but, as previously shown, the separation of rn-chlorobenzoic and p -

+

chlorobenzoic acids then becomes the limiting factor. The slight-peak tailing for hydroxybenzoic acids may be attributed to a partial thermal decomposition which occurs in the chromatographic apparatus. CONCLUSION The experimental data here reported make it clear that GLS chromatography constitutes an effective tool for solving a number of analytical problems which are hard to solve with conventional gas chromatographic techniques. Suitable adsorptive modifications in the graphitized carbon blacks make it possible to exploit the advantages of adsorption chromatography, the most important being the high selectivity toward molecules differing mainly in their geometric structure. It also permits the use of high linear gas velocities without loss of performance. At the same time, the use of a small amount qf an involatile liquid added to the adsorbing materials eliminates the initial "Knee-bend'' in the adsorption isotherms for polar compounds. Furthermore, the cooperative effects of the liquid and the adsorbing surface can enhance, in some cases, the selectivity of a GLS column. Final!y, it has to be pointed out that the volatility of monolayers deposited on an adsorbing surface is to a varying degree lower than that of the corresponding bulk liquid. Therefore, bleeding of the stationary phase can be attenuated with the use of GLS columns. ACKNOWLEDGMENT The author is indebted to A. Liberti and to R . Samperi for helpful discussions, and to N. Neri for technical assistance. RECEIVED for review June 30, 1972. Accepted November 1, 1972.

Separation of Polya romatic Hyd roca rbons by Liquid-Sol id Chromatography Using 2,4,7=Trinitrofluorenone Impregnated Corasil I Columns B. L. Karger,l M. Martin, J. Loheac, and G . Guiochon Ecole Polytechnique, Laboratoire de Chimie Analytique Physique, 17, rue Descartes, Paris V " , France

The use of trinitrofluorenone as stationary phase in liquid chromatography has been investigated. Very specific interactions occur so that many isomers of polynuclear aromatic hydrocarbons with 5 and 6 rings have been resolved. Corasil I Dorous laver beads used as support contribute partly 'to the retention. This effect can be reduced by'using water saturated heptane as the mobile phase. The retention times are then markedly reduced.

complexation to €Thrice selectivity in chromatography is well known. Silver nitrate and silver oxide doped liquids and adsorbents have been employed for many years for the selective retardation of compounds containing olefinic double bonds via charge THE

USE OF CHARGE TRANSFER

On sabbatical leave 1/72-4/72; permanent address, Department of Chemistry, Northeastern University, Boston, Mass. 496

ANALYTICAL CHEMISTRY, VOL. 45, NO. 3, MARCH 1973

transfer comolexation between the silver ion and the unsaturated center. Another useful acceptor is 2,4,7-trinitrofluorenone (TNF) which has found application in gas chromatography (1-4) and thin layer chromatography (5, 6). This substance is - . especially selective to aromatics with three fused rings or greater, for in such cases orbital overlap is favorable (7). A

(1) R. 0. C.

Norman, Proc. Chern. Soc., 1958, 151. (2) A. R. Cooper, C. W. P. Crowne, and P. G. Farrell, Trans. Furuday Soc., 62, 2725 (1966). (3) Ibid., 63,447 (1967). (4) c. w. p. CrOwne, M. F. Harper, and p. G. Farrell, J . togr., 61,7(1971).

( 5 ) A, Berg and J. Lam, ibid,, 16, 157 (1964), (6) R. G. Harvey and M. Halonen, ibid., 25, 294 (1966). (7) G. Schenk, P. Vance, J. Pietrandrea, and C. Mojzis, ANAL.

CHEM.,31,3'72 (1965).

Good separations of aromatic substances were notably obtained by Harvey and Halonen ( 6 ) via different strengths of charge transfer complexation with TNF. That a charge transfer complex indeed forms was strongly inferred in this work by the linear relationship of the formation constants t o R F values. The emergence of column liquid chromatography over the last few years as a viable separation tool has been well publicized (8, 9). Speeds and efficiencies approaching packed gas chromatographic columns are now possible. Part of the reason for this development has been the introduction of special packing materials, namely porous layer beads (PLB) [ e . g . , Zipax (IO), Corasil ( I ] ) ] . These materials offer rapid mass transfer possibilities due to the small pore depths resulting from the incorporation of a thin porous layer atop a fluid impenetrable spherical core. The purpose of this paper is to illustrate the combination of selective separations via charge transfer complexation with the speed and efficiency possible from PLB. Separation of aromatic hydrocarbons using T N F impregnated Corasil I columns will be examined. To our knowledge there has been only one other report on the use of complexation with PLB, namely silver oxide impregnated on Zipax (12).

Table I. Relative Change Retention on TNF Impregnated to Corasil I Columns, T = 25 “C Solute

e ia; 3.05 Z T N F 0.10 .67

Anthracene

% TNF

4 .O

Pyrene

1.9

6 .I

Chrysene

2.2

4 .O

34,

I&

- benzpyrene

4.1

11 .6

4.3

10.5

9.9

(e.0) (b) k; .capacity ratio on Corasil

I alone

k; .capacity rotio on TNF Impregnated Corasil I (b)

EXPERIMENTAL

Toiling was bad ror this compound on both the Comsil

0.05

Apparatus. The home-made apparatus has been previously described in detail (13). To summarize briefly, this consisted of a n Orlita pump M 3 S4/4, thermostated column, pressure regulator (149,and LKB ultraviolet detector with a specially designed cell. For low inlet pressures, a simple Hamilton 10-pl syringe was used; however, a t high pressures, a Hamilton 5-pl high pressure syringe HP was employed. Column Packing. The packing material was Corasil I (37-50 pm) (Waters Associates). Before coating with T N F (Schuchardt, Munich), the adsorbent was dried for 3 hours at 140 “C. TNF (of the required weight) was dissolved in acetone (dried with molecular sieve 3A) and then the solution was mixed with the adsorbent in a rotary evaporator. The acetone was removed by applying a vacuum with stirring. T o remove the final traces of acetone, the packing was heated for 2-3 hours a t 50-70 “C. The column (1.6-mm i.d. stainless steel) was filled vertically by small additions of the dry packing. After each increment, the column was tapped with a rod and the support of the column was vibrated. (The column itself was not vibrated in order to avoid nonuniform packing densities.) After filling, the column was tapped on the ground to help settle the packing and to permit the addition of more support. The packing was maintained in the column by small wire gauze. The 3-meter column was made up in 3 sections of one meter each. The sections, after filling, were joined by Swagelok unions and tubing (2-mm i.d.; 3 m m long). The tubing connections were filled with packing material. Since coiling with a small radius of curvature results in loss in efficiency (IS), (8) “Modern Practice of Liquid Chromatography,’’ J. J. Kirkland, Ed., Interscience New York, N.Y., 1971. (9) J. F. K. Huber, J . Chromatogr. Sci., 7, 85 (1969). (10) J. J. Kirkland, ibid., p 7. (11) J. N. Little, D. F. Horgan, and K. J. Bornbaugh, ibid., 8, 625 (1970). (12) R. Vivilecchia and R. Frei, ibfd., 10,411 (1972). (13) M. Martin, J. Loheac, and G. Guiochon, Clzromatographia, 5 , 3 3 (1972). (14) G. Deininger and I. Halasz, J. Chromatogr., 60, 65 (1971). (15) H. Barth, E. Dallrneier, and B. L. Karger, ANAL.CHEM.,44, 1726 (1972).

Structure

I and

z TNF columns

the column was made into a U shape with straight lengths of 1.5 meters each. This column was not thermostated. The mobile phase in this work was n-heptane dried over activated molecular sieves. In some experiments, heptane saturated with water was employed. RESULTS AND DISCUSSION Before presenting the results, we wish to point out a technical problem. As the number of fused aromatic rings increases, the maximum of absorbance in the ultraviolet region shifts to longer wavelengths. Thus, while 254 nm H g light is quite good for 1-, 2-, or 3-ring aromatics, sensitivity becomes a problem especially with 5- and 6-fused ring systems. A larger sample is necessary to be able to detect these latter compounds a t 254 nm, resulting in column overloading and sample size dependent retention. The obvious solution to this problem is to use a different wavelength for detection (e.g., 280 nm), but unfortunately a detector with this capability was not available for this investigation. We have attempted to minimize this overloading effect by using sample sizes as small as possible. While peak tailing is not excessive, the retention parameters for 5- and 6-fused rings should be viewed as qualitative indication of adsorption, not quantitative. (A second cause of peak asymmetry, namely kinetics of the adsorption process, will be discussed later.) From some preliminary experiments on silica gel, it was estimated that roughly 0.05 % T N F o n Corasil I would result i n reasonable retention and efficiency for 4- and 5-ring fused aromatics. We therefore made two 50-cm columns of 0.05 wjw T N F and 0.10% w/w T N F on Corasil I and compared retention to a column of Corasil I without T N F , using in all cases dry heptane as the mobile phase. Independent tests revealed no solution of T N F into heptane from the Corasil I support. The results of this comparison are shown in Table I. In this table, we report the relative change in capacity factor of a solute from “bare” Corasil I to a column coated with

ANALYTICAL CHEMISTRY, VOL. 45, NO. 3, M A R C H 1973

497

2.5

-

2

-

Table 11. Influence of Water Deactivation on Retention (Capacity Factor) T = 24 “C 0.05% TNF 0.10% TNF

k’, 0.29 1

k,O ’

0.5



0

I

A-

k’=O k‘=O.10

HepSolute tane Anthracene 0.13 Pyrene 0.29 Chrysene 0.64 3,4-Benzpyrene 1.4 1,2,3,4-Dibenzanthracene 2 . 5 1,2,3,4-Dibenzpyrene 7.1 (7.2) 3,4,8,9-Dibenzpyrene

Hep-

Hep-

tanewatersatd 0.15 0.18 0.33 0.79 1.2 2.9 3.6

tanewater-

Heptane 0.40 0.71 1.4 3.5

5.4 large large

satd

0.09 0.23 0.46 1.2 1.9 4.8 large

Table 111. Influence of Temperature on Retention and Selectivity TNF on Corasil I Column-0.05 T = 24 “C T = 58 “C Solute k‘ cr k; (Y 0.14 0.11 Anthracene 0.29 0.20 Pyrene 0.30 Fluoranthene Chrysene 2.0 3,4-Benzpyrene 1,2,3,4-Dibenzanthracene 1,2,3,4-Dibenzpyrene

1

2

3 u(cm/s)

3,4,8,9-Dibenzpyrene

(7.2)

3.4

Figure 1. Variation of HETP us. flow velocity (dry n-heptane). Temperature: 24 “C, L = 50cm A : 0.25 TNF impregnated on Corasil I (37-50,~)k‘ = 0 benzene: k ’ = 0.29 pyrene; k‘ = 0.64 chrysene B: uncoated Corasil I (37-50 H). k‘ = 0 benzene; k‘ = 0.10

pyrene

TNF. The parameter, 8, is then a qualitative indicator of the influence of T N F on retention. We first note that for a given T N F column, the change in retention, 8, is a function of the number of fused aromatic rings in the system. Thus on the 0.05 % T N F column, 8 = 0.7 for 3 rings, 2 for 4 rings, 4 for 5 rings, and 8-10 for 6 rings. These results are in the expected direction, as the complex between aromatics and T N F is known to be stronger as the number of fused rings is increased. Stated differently, the trends suggest that a charge transfer complex is forming. The results further indicate that from an analytical point of view, separation of aromatics into classes dependent on the number of fused rings is a simple matter. An additional indication of the influence of T N F is shown in the increase in 8 for a given solute from the 0.05% to the 0.10% impregnated column. Yet a strict 2 : l relationship is not found for 8 with the two columns. As we will shortly show, both Corasil I and T N F contribute to retention, as Corasil has been shown to possess an active, heterogeneous surface (16). For 6-ring systems, retention was too long on the 0.10% column to provide meaningful k’ values to report in Table I. (16) B. L. Karger, H. Engelhardt, K . Conroe, and I. Halasz, “Gas Chromatography, 1970,” R. Stock, Ed., Institute of Petroleum, London, 1971. 498

ANALYTICAL CHEMISTRY, VOL. 45, NO. 3 , MARCH 1973

Table I1 presents the influence of water deactivation of the two T N F impregnated columns. For this experiment, each column was first tested with dry heptane as the mobile phase. Then heptane saturated with water was allowed to flow through the column overnight and once a steady state was achieved, measurements were made with this mobile phase. A significant drop in retention occurs when the adsorbent is deactivated with water. The results point strongly to the participation of bare Corasil I adsorption sites to the retention of the aromatic hydrocarbons when dry heptane is used. I t is also possible that the complexation constant could change with the change in solvent; however, the predominant effect should be the deactivation of Corasil. Figure 1 presents H us. velocity plots for Corasil I alone and 0.05% T N F impregnated on Corasil I with dry heptane as mobile phase. For unretained components efficiencies comparable to those published in the literature for Corasil I (15,16) are attained. As k’ for solutes increases, the efficiency at any given velocity decreases. For solutes with k’ above roughly 3.0 on the 0.05 % T N F column, there i s a precipitous drop in efficiency (down to 300 plates per meter or less at u = 1.0 cmjsec). This large drop is due in part to the asymmetrical peak that is obtained as a result of column overloading. We would also anticipate a drop in efficiency as the formation constant for complexation (and thus k ’ ) increases because the rates of the complexation and dissociation are known to decrease. Thus, mass transfer from the mobile to the stationary phase and back is slowed down. Beside column overloading, the peak asymmetry may arise from two other sources. First, if the rate constants for formation and dissociation of the charge transfer complex differ, the

, 0

10

I

I

20

30

1

40

t (mnl

-

1

Figure 2. Separation of 8 polyaromatic hydrocarbons whose structures are indicated in the diagram. Column: 3 meters, 1.6. mm i.d.; 0.06% TNF on Corasil I(37050 pm; mobile phase, dry heptane; T = 25 "C, Ap = 185 atm, u = 2.1 cm/sec

Table IV. Retention of Additional Solutes on TNFCorasil I. Mobile Phase-Dry Heptane, T = 24 "C 0.05 %TNF 0.10 % TNF Phenanthrene

0.14

Fluarant hene

0.24

0.67

1,2- benzanthracene

0.56

1.2 3.1

1,2-benzpyrene

Figure 3. Separation of 1,2-benzpyrene and 3,4-benzpyrene. Same column and conditions as Figure 2 except A p = 80 atm, c = 0.9 cm/sec rates of mass transfer between the stationary and mobile phases would also differ, resulting in peak asymmetry. Second, as we have previously shown, retention occurs via adsorption of the Ijolyaromatic hydrocarbons on T N F and base Corasil I. Such a heterogeneous surface could also give rise to peak asymmetry. Returning to Table 11, besides the drop in k' for each solute, we find an increase in the number of theoretical plates. For example at u = 1 cmjsec the theoretical plates on the 0.05% T N F column of 50 cm are 670 and 930 (dry heptane and heptane/H20, respectively) for the solute pyrene, and 150 and 230 (dry heptane and heptane/H?O) for the solute 1,2,3,4dibenzanthracene. Thus one needs to balance improved efficiency with optimum capacity factors ( k ' ~ 1 - 3 )for minimum analysis time. In effect, the water deactivation converts the column from one best suited to 4- and 5-ring compounds to one suited to 5- and 6-ring compounds as seen in Table 11. If the drop in efficiency is caused by slow kinetics of complexation/dissociation, then an increase in column temperature might prove useful. In addition, liquid viscosity is known to

3.7

Perylene

1.4

2,3,6.7-dibenzanthracene

1.9

1,2,5,6- dl benzanthracene

2.2

5.2

1,2,7,8-di benzphenanthrenl

2.5

7.0

decrease with rise in temperature, resulting in better mass transfer in the mobile phase and a lower pressure drop for a given velocity. We have examined the 0.05% T N F column (dry heptane as mobile phase) at T = 24 "C and 58 "C and find roughly a 20% increase in theoretical plates. As seen in Table I11 there is also a decrease in k' and, as expected, a decrease in relative retention. The gains in efficiency appear to be offset by the loss in selectivity; however, temperature can be a useful parameter in controlling k' values of the polyaromatics. The compromise between selectivity and efficiency is further seen in a comparison of the 0.05% and 0.10% T N F columns ANALYTICAL CHEMISTRY, VOL. 45, NO. 3, MARCH 1973

499

Figure 4. Separation of 1,2,3,4-dibenzpyrene and 3,4,8,9-dibenzpyrene. Column is same as in Figure 2. Mobile phase = heptane saturated with water. c = 1.9 cmisec (dry heptane as mobile phase). For any given compound, only two thirds as many theoretical plates are generated on the heavier impregnated column, but in general a is found to be larger, especially in the separation of compounds with different ring numbers. In the applications section of the paper, we will present a 3-meter column made up of 0.06% T N F which strikes a compromise between efficiency and retention/selectivity. In Table IV, we present the retention of other fused ring aromatic compounds that we have run on the 0.05 and/or 0.10% T N F columns. As in the other cases, retention is seen to increase with the number of rings in the molecule. Note also the selectivity for solutes of the same number of fused rings.

Figure 2 presents the separation of 8 aromatic hydrocarbons at c = 2.1 cmjsec. As noted previously, the bands appear in groups, depending on the number of fused aromatic rings. This column with dry heptane as the mobile phase is most suited to 4- and 5-ring polyaromatics. As shown in Figure 3, the two benzpyrene isomers are completely resolved at a lower velocity ( c = 0.9 cmjsec). This separation has been found to be difficult by gas-liquid chromatography. By switching the mobile phase to heptane saturated with water, we deactivate Corasil I adsorption sites, thus decreasing k‘ values for all the solutes. In this case the 3-meter column becomes well suited to the separation of 6ring systems. Figure 4 shows an example of this in the separation of two isomers of dibenzpyrene at c = l .9 cmjsec.

APPLICATIONS

CONCLUSION

Having examined some of the characteristics of T N F , impregnated Corasil I column, we will now show some typical applications. For this purpose we built a 3-meter column containing 0.06% T N F on Corasil I. We chose the long column to provide sufficient theoretical plates at reasonable velocities for separation. The permeability of the column was such that only 180 atm was required to generate a velocity of 2 cmjsec. For benzene (k’ = 0) over 7,000 theoretical plates were achieved at c = 0.3 cmjsec. Even for retained components, several thousand theoretical plates were attained at this velocity. (While efficiency decreases at higher velocities, the chromatograms, Figures 2-4, show that relatively good efficiencies are still achieved.) We chose 0.06 loading as a reasonable compromise between 0.05% and 0.10% T N F in terms of efficiency and selectivity.

We have shown in this paper that TNF-impregnated PLB columns are useful tools in the separation of polyaromatic compounds. By adjustment of % TNF, mobile phase water content, and temperature, it is possible to optimize the column for rapid analysis (i.e., k ’ near 2 ) for the separation of 3-, 4-, 5-, 6-, or even 7-fused ring polyaromatic systems. Clearly, there are many other potential applications of these columns. Especially interesting would be the study of the separation of heterocyclics using T N F , as successful separations have been achieved in gas-liquid chromatography (2-4).

500

ANALYTICAL CHEMISTRY, VOL. 45, NO. 3, MARCH 1973

RECEIVED for review May 30, 1972. Accepted October 30, 1972. B. L. Karger is a Fellow of the Alfred E. Sloan Foundation, 1971-73. Work partially supported by NIH under Grant GM-15847.