Resolution of Small Molecules by Passage through an Open Capillary

Aug 17, 2005 - When a solute passes through an open capillary, in which a laminar flow is established, different peak profiles can be obtained accordi...
0 downloads 10 Views 145KB Size
Download PDF
Anal. Chem. 2005, 77, 6041-6046

Resolution of Small Molecules by Passage through an Open Capillary Tetsuo Okada,* Makoto Harada, and Tomoo Kido

Department of Chemistry, Tokyo Institute of Technology, Meguro-ku, Tokyo 152-8551, Japan

When a solute passes through an open capillary, in which a laminar flow is established, different peak profiles can be obtained according to its diffusion property under a working condition, i.e., the radius and length of the capillary and the flow rate of the carrier solution. If a solute diffuses over the entire cross section of the capillary before it is eluted, a Gaussian-shaped diffusion peak appears, which has an apex at the travel time of the average flow. Insufficient solute diffusion, which is realized, e.g., by increasing flow rates or capillary radius, produces a new peak having an apex at the travel time of the maximum flow. This implies that two solutes can be resolved simply by passing through a capillary. However, our previous study indicated that the diffusion coefficients of two solutes should be at least one order different for their resolution based on this principle, suggesting that its applicability is highly restricted. In the present paper, this concept has been extended to the resolution between dissolved solutes that have similar intrinsic diffusion properties. The incorporation of molecular aggregates in the carrier makes a solute less diffusive according to the extent of their interaction and allows the resolution of a dissolved molecule from other ones differing in the affinity to the molecular aggregates. Several examples of peak resolution for phenols, aromatic hydrocarbons, and inorganic anions are shown and discussed on the basis of the modification of the diffusion natures due to their interactions with micelles or vesicles. Chromatography, electrophoresis, and field flow fractionation are employed to separate various classes of solutes involving simple molecules, macromolecules, particles, etc. One of the recent trends of separation science is the miniaturization of systems, called lab-on-chip or µ-TAS, which has been strongly supported by the developments of microfluidics and micromechanics.1-5 Another important direction is to develop novel separation principles using physical or chemical fields. Although physical external fields, such as magnetic, optical, and acoustic fields, have recently been shown to be useful in some applications, these force * To whom correspondence should be addressed. Phone and Fax: +81-35734-2612. E-mail: [email protected]. (1) Gardeniers, J. G. E.; van den Berg, A. Bioanal. Chem. 2004, 378, 1700. (2) McCreedy, T. Anal. Chim. Acta 2001, 427, 39. (3) Uchiyama, K.; Tokeshi, J.; Kikutani, Y.; Hattori, A.; Kitamori, T. Anal. Sci. 2005, 21, 49. (4) Shintani, T.; Torimura, M.; Sato, H.; Manabe, T. Anal. Sci. 2005, 21, 57. (5) Petersson, F.; Nilsson, A.; Jo ¨nsson, H.; Laurell, T. Anal. Chem. 2005, 77, 1216. 10.1021/ac050715q CCC: $30.25 Published on Web 08/17/2005

© 2005 American Chemical Society

fields have not been well utilized because special (mostly ordermade) instruments are required, and their general application is therefore restricted to some extent.5-9 New separation fields are more extensively studied from chemical viewpoints. Chromatographers, for example, are still exploiting new stationary phases, which allow separation not performed with conventional materials.10-12 Molecular aggregates, which usually provide hydrophobic environments as well as electrostatic interactions, have played important roles in the developments of novel chemical separation fields. The incorporation of micelles in chromatographic mobile phases led to the development of micellar chromatography,13 and their utilization in electrophoresis allowed the invention electrokinetic chromatography.14 Micelles obviously have advantages over other molecular aggregates, such as vesicles and microemulsions, because they provide homogeneous and optically transparent solutions and thus are easily applicable to various analytical systems. For these reasons, uses of vesicles and microemulsions in separation are relatively fewer than those of micelles.15-18 Separation fields have thus been devised using chemical interactions or physical external fields. In contrast, hydrodynamic chromatography is known as a slightly different method, in which such forces or interactions are not explicitly involved.19-22 Particles carried by a flow cannot completely approach the channel wall because of the existence of the exclusion volume, which is a function of the particle size. Particles are thus separated according (6) Masudo T.; Okada T. Anal. Chem. 2001, 73, 3467. (7) Masudo T.; Okada T. Anal. Sci. 2004, 20, 753. (8) Watarai, H.; Monjushiro, H.; Tsukahara, S.; Suwa, M.; Iiguni, Y. Anal. Sci. 2004, 20, 423. (9) Kaneta, T.; Ishidzu, Y.; Mishima, N.; Imasaka, T. Anal. Chem. 1997, 69, 2701. (10) Li, L.-S.; Da, S.-L.; Feng, Y.-Q.; Liu, M. J. Chromatogr., A 2004, 1040, 53. (11) Kim, H.; Guiochon, G. Anal. Chem. 2005, 77, 93. (12) Huang. X.; Wang, J.; Wang, Q.; Huang, B. Anal. Sci. 2005, 21, 253. (13) Armstrong, D. W.; Nome, F. Anal. Chem. 1981, 53, 1662. (14) Terabe, S.; Otsuka, K.; Ando, T. Anal. Chem. 1985, 57, 834 (15) Delgado-Zamarren ˜o, M. M.; Sa´nchez-Pe´rez, A.; Maza, I. G.; Herna´ndezMe´ndez, J. J. Chromatogr., A 2000, 871, 403. (16) Razak, J.L.; Cutak, B. J.; Larive, C.K.; Lunte, C. E. Pharm. Res. 2001, 18, 104. (17) Pysher, M.; Hayes, M. A. Langmuir 2004, 20, 4369. (18) Furumoto, T.; Fukumoto, T.; Sekiguchi, M.; Sugiyama, T.; Watarai, H. Electrophoresis 2001, 22, 3438 (19) Mullins M. E.; Orr, C. Int. J. Multiphase Flow 1979, 5, 79. (20) Bos, J.; Tijssen, R.; van Kreveld, M. E. Anal. Chem. 1989, 61, 1318. (21) Chmela, E.; Tijssen, R.; Blom, M. T.; Gardeniers, H. J. G. E.; van den Berg, A. Anal. Chem. 2002, 74, 3470. (22) Blom, M. T.; Chmela, E.; Oosterbroek, R. E.; Tijssen, R.; van den Berg, A. Anal. Chem. 2003, 75, 6761.

Analytical Chemistry, Vol. 77, No. 18, September 15, 2005 6041

to their sizes just by passing through a channel. However, if the channel height (or capillary radius) is much larger than the particle size, this mechanism does not work for solute resolution because the exclusion volume is negligibly small in comparison with the entire volume of the separation channel. In contrast, a completely different mechanism originating from the solute distribution coupled with the laminar flow profile should be taken into account, when solutes flow through an open channel having large dimension; separation based on this mechanism has been named wide-bore hydrodynamic chromatography.23,24 A diffusive solute gives an elution curve with an apex at the time required for the average flow to reach the detector, whereas a nondiffusive solute gives an asymmetric elution curve having an apex at the travel time of the maximum flow, which is twice as high as the average flow rate in a capillary.24 The numerical simulations based on the diffusion-advection equation has shown that two solutes can be resolved simply by passing through an open capillary if their diffusion constants differ by 1 order of magnitude. The diffusivity of a solute strongly depends on the system and flow conditions. It is well known that the average square displacement (x2) is given by

x2 ) 2Dt

(1)

where D is the diffusion coefficient of a solute. The time required for a solute traveling with the maximum flow to pass through an open capillary (a is radius, and L is length) should be equal to t ) L/(2uav), where uav is the average linear flow rate. The following condition should thus be satisfied for all of the solute to diffuse over the entire capillary cross section before it is eluted from the capillary:

a2 e 2Dt ) DL/uav τav ) DL/(a2uav) g 1

(2)

This nondimensional time (τav) can also be represented conveniently by the volume flow rate (F/m3 s-1),

τav ) πDL/F Thus, τav is a very convenient parameter to discuss the solute diffusivity; if τav g 1, it is diffusive under a given condition and gives a Gaussian peak, but if τav < 1, it does not completely diffuse and produces a non-Gaussian peak. This parameter implies that the diffusivity of a solute can be varied by changing a, L, and uav (or F) without modifying D. According to our simulation, a nondiffusive peak is seen if τav < 0.1, and an intermediate one is observed if 0.1 < τav < 1. Thus, the diffusion coefficient of a given solute should be 10 times or one-tenth as large as that of another solute for complete resolution. Since small molecules dissolved in water generally have D ) 10-9-10-10 m2 s-1, their mutual resolution based on this principle is not possible. However, if molecular aggregates, such as micelles and vesicles, are incorporated in a carrier in wide-bore capillary hydrodynamic chro(23) Fischer Ch.-H.; Giersig, M. J. Chromatogr., A 1994, 688, 97. (24) Harada, M.; Kido, T.; Masudo, T.; Okada, T. Anal. Sci. 2005, 21, 491.

6042

Analytical Chemistry, Vol. 77, No. 18, September 15, 2005

matography, the diffusivity of a solute should be reduced according to the extent of the interaction with them. In the previous paper, we showed that SDS micelles can be resolved from dissolved molecules not partitioned into the micelles.24 This separation principle can be extended to the resolution of dissolved molecules if appropriate molecular aggregates are incorporated in the carrier of wide-bore capillary hydrodynamic chromatography. In this paper, we present the first experimental evidence for this concept and discuss its methodological potentials and limitations. EXPERIMENTAL SECTION A fused-silica capillary of 320-µm i.d. as well as a glass capillary of 1-mm i.d. was used for flow experiments. The effective lengths of the capillaries were ∼30 cm. The mobile phases were fed by a syringe pump model 11 (Harvard) equipped with a 1- or 10-mL gastight syringe. Sample solutions were introduced with a Rheodyne HPLC injector; the injection volume of the sample (0.5 µL for 320-µm capillary and 5 µL for 1-mm glass tube) was controlled by the activation time of the injector. The elution of a solute from the capillary was monitored by a UV/visible detector 870-CE (Jasco). The full scale of the detector was set to 0.04 AU. The absorption was measured for the central part of the capillary using a 50-µm slit. The sizes of vesicles were measured by light scattering with a Wyatt model DAWN EOS. SDS was purified by recrystallization. Vesicles were prepared as follows: an appropriate aliquot of methanolic solution of dialkyldimethylammonium bromide was added into ∼40 mL of boiling water; after cooling, the resulting solution was diluted in a 50-mL volumetric flask. Three dialkyldimethylammonium bromides were used for vesicle preparation, i.e., ditetradecyldimethylammonium bromide (TMA), dihexadecyldimethylammonium bromide (HMA), and dioctadecyldimethylammonium bromide (OMA). Phenols and aromatic hydrocarbons were dissolved in methanol or acetonitrile, and then their small portions (typically 5-10 µL) were added into 10 mL of the carrier solution containing micelles or vesicles. Pyrene was used as a diffusion marker for the SDS micelle and vesicles because of its extremely high binding ability to these molecular aggregates. The concentration of pyrene in sample solutions was kept as low as 0.025 mM. The concentrations of other solutes were 0.05-0.2 mM, depending on their absorptivity. MilliQ water was used for solution preparation. The diffusion coefficients of molecular aggregates were determined by the method reported in our previous paper.24 RESULTS AND DISCUSSION Elution of Micelles and Vesicles from an Open Capillary. There are two essential factors to separate small diffusive molecules by wide-bore hydrodynamic chromatography with carrier solutions containing molecular aggregates: (1) there should be an appropriate difference in the solute affinity to the molecular aggregates; (2) molecular aggregates behave as nondiffusive solutes. The reported diffusion coefficients of SDS micelles, which depend on the concentration, are summarized in Table 1. In the present study, the length of the capillary was kept 0.3 m irrespective of the capillary radius. Thus, a2uav () DL/τav) should be as large as 1.9 × 10-10 for 100 mM SDS (D ) 6.43 ×

Table 1. Diffusion Coefficients of Selected Solutes solute

concentration

D/m2 s-1

Iphenol acetone SDS micelle

limiting dilution limiting dilution limiting dilution 50 mM 100 mM

2.045 × 10-9 a 8.9× 10-10 b 2.17 × 10-9 a 7.52 × 10-11 c 6.43 × 10-11c

TMA vesicle HMA vesicle OMA vesicle

radius/nm

D/m2 s-1 d

D/m2 s-1 e

311 ( 54 52.9 ( 1.7 39.8 ( 3.8

1.36 × 10-12 8.28 × 10-12 1.10 × 10-11

1.96 × 10-12 9.81 × 10-12 1.96 × 10-11

a Taken from Kagaku Binran (Chemical Index); Maruzen: Tokyo, 2004; Version 5. b Taken from ref 32. c Taken from ref 25. d Calculated from the radius of the vesicles based on the Stokes law. e Determined by the method reported in ref 24.

10-11 m s-1);25 i.e., uav ) 7.6 × 10-4 m s-1 for 1-mm-i.d. capillary and 7.4 × 10-3 m s-1 for 320-µm-i.d. capillary. The former capillary was selected for SDS because the smaller linear flow rate is desirable to keep the laminar flow profile (to maintain smaller Reynolds number). It is well known that the size of vesicles strongly depends on the methods used for their preparation. Chloroform injection, where surfactants dissolved in this solvent are injected into water, usually gives large vesicles. A very large one called a gigantic vesicle has also been prepared when desired.26-29 In contrast, it is known that the addition of alcoholic solution into hot water yields relatively small vesicles; for example, the radius of the OMA vesicle prepared by this method is reported to be some tens of nanometers.27 In the present study, three types of double-tailed ammonium ions were examined to prepare vesicles. TMA (10 mM) gave turbid solution, whereas HMA and OMA (10 mM) resulted in transparent solutions, suggesting that the vesicular sizes depend on the chain lengths of the ammonium ions. The vesicular sizes determined by a light scattering are listed in Table 1. As expected from the transparency of the solutions, the vesicular size decreases with increasing alkyl chain length of the doubletailed ammonium salts. The sizes of the vesicles allow the estimation of D on the basis of the Stokes law as also listed in Table 1. We showed that the diffusion coefficients can be evaluated with wide-bore capillary hydrodynamic chromatography and that this method is advantageous for the determination of very small values, e.g., those for vesicles.24 Increasing the flow rate (uav) in widebore capillary hydrodynamic chromatography caused a solute to be less diffusive, leading to the development of a nondiffusion peak. Although the front edge position of an elution curve continuously changes with uav for a diffusive solute, the transformation of a diffusion peak to a nondiffusion peak results in an abrupt shift of the flow rate dependence of the front edge position; (25) Davis, J. M. Analyst 1998, 123, 337. (26) Deamer, D.W.; Uster, P.S. In Liposomes; Ostro, M. J., Ed.; Marcel Dekker: New York, 1983; Chapter 1. (27) Kuwamuro, M. K.; Chaimovich, H.; Abuin, E. B.; Lissi, E. A.; Cuccovia, I. M. J. Phys. Chem. 1991, 95, 1458. (28) Cuccovia, I. M.; Feitosa, E.; Chaimovich, H.; Sepulveda, L.; Reed, W. J. Phys. Chem. 1990, 94, 3722. (29) Blandamer, M. J.; Briggs, B.; Cullis, P. M.; Kirby, S. D.; Engberts, J. B. F. N. J. Chem. Soc., Faraday Trans. 1997, 93, 453.

Figure 1. Elution profiles of vesicles. (A) Effect of flow rate on the elution of HMA vesicle: average linear flow rate (uav), (1) 8.30 × 10-4 (4 µL min-1), (2) 4.14 × 10-3 (20 µL min-1), and (3) 1.04 × 10-2 m s-1 (50 µL min-1). (B) Elution curves of different vesicles: average linear flow rate, 4.14 × 10-3 m s-1 (20 µL min-1). Detection wavelength, 260 nm.

the front edge position of the nondiffusion peak is almost independent of the flow rate. Thus, when the front edge of the elution curve is plotted against uav1/2 (the Gaussian function predicts linear changes in the front edge with uav1/2), a flexion occurs at a particular uav1/2, which corresponds to the time required for a solute traveling with the maximum flow to diffuse over the cross section of r ) 0.86a.24 The values determined for vesicles are also listed in Table 1. Figure 1A depicts the deformation of elution curves of the HMA vesicle with changing flow rates. As can be seen in eq 1, increasing flow rates results in a decrease in τav and in turn makes the solute less diffusive. This figure obviously indicates that a nondiffusion peak grows with increasing flow rate. Figure 1B compares the elution curves obtained for different vesicles at the same linear flow rate (uav ) 4.14 × 10-4 m s-1). A diffusion peak appears for the OMA vesicle, whereas a large nondiffusion peak is seen for the TMA vesicle; the HMA vesicle shows an intermediate characteristics. The difference in the peak shape is obviously explained by different diffusion natures of these vesicles. The τav values are 0.055, 0.28, and 0.55 for the OMA, HMA, and TMA vesicles under this condition, respectively, indicating that even the OMA vesicle is not completely diffusive under this condition. Wide-Bore Capillary Hydrodynamic Chromatography of Selected Compounds. The apparent diffusion coefficient of a Analytical Chemistry, Vol. 77, No. 18, September 15, 2005

6043

Figure 2. Calculated relations between log R and uav. Curves 1, 2, and 3 indicate the boundaries between nondiffusive and intermediate peaks, while curves 1′, 2′, and 3′ refer to the boundaries between intermediate and diffusive curves, respectively. D2, (1, 1′) 2 × 10-9, (2, 2′) 1 × 10-9, and (3, 3′) 5 × 10-10 m2 s-1. Three horizontal lines show log R values for phenols in a 100 mM SDS system.

solute (Dapp) is a function of the extent of the solute interaction with the molecular aggregates incorporated in the carrier solution and should range from that of the molecular aggregate to that of the solute itself. If we attempt to resolve two solutes based on the present concept, one of the solutes should be nondiffusive (τav < 0.1) under the condition where the other is diffusive (τav > 1). The apparent diffusion coefficients of these solutes should thus be at least one order different. The apparent diffusion coefficient of a solute interacting with a molecular aggregate may be given by the linear combination of two diffusion coefficients, i.e.

Dapp ) RD1 + (1 - R)D2

(3)

where D1 and D2 are the diffusion coefficients of the molecular aggregate and the solute in a pure solvent, respectively, and R is the degree of solute binding. Dapp can be calculated assuming appropriate D1 and D2. As stated above, a diffusion peak is obtained if τav is larger than unity, whereas a complete nondiffusion peak is seen when τav is smaller than 0.1; an intermediate peak is given for τav ) 0.1-1. We can thus predict peak profiles with R and uav as experimentally variables for a given dimension of the separation capillary. Figure 2 predicts peak profiles with varying log R and uav for the SDS micellar system in a 1-mm-i.d. capillary and the HMA vesicular system in a 320-µm-i.d. capillary. As expected from eq 3, D2 rather than D1 strongly influences resulting Dapp values; D2 is varied 5 × 10-10 -2 × 10-9 m2 s-1 in Figure 2. In contrast, 6044 Analytical Chemistry, Vol. 77, No. 18, September 15, 2005

Figure 3. Elution profiles of phenolic compounds. Carrier solution, 100 mM SDS. Capillary, 1 mm i.d. × 0.3 m. Average linear flow rate, (1) 2.12 × 10-4 (10 µL min-1), (2) 4.25 × 10-4 (20 µL min-1), (3) 8.49 × 10-4 (40 µL min-1), (4) 1.27 × 10-3 (60 µL min-1), and (5) 1.70 × 10-3 m s-1 (80 µL min-1). Detection wavelength, 270 nm.

changes in D1 resulted in very marginal differences, albeit the results are not shown. Increasing uav transforms a diffusion peak of a given solute into a nondiffusion peak through an intermediate peak profile. Alternatively, increasing R gives the same transformation of peak shapes; R can be actually modified by changing types or concentration of molecular aggregates. Figure 2 also indicates that a wider bore capillary facilitates nondiffusive characteristics of less partitioned solutes. Thus, the wider bore provides higher resolution capability but is not suitable to maintain a laminar flow profile as mentioned already. Figure 3 compares elution curves of phenol, p-chlorophenol, and naphthol with 100 mM SDS as the carrier solution. Phenol does not behave as a perfect nondiffusive solute within the range

Figure 4. Resolution between naphthol and phenol with 100 mM SDS carrier. Average linear flow rate, 8.49 × 10-4 m s-1 (40 µL min-1). Other conditions are the same as Figure 3.

Figure 5. Resolution between anthracen and naphthalene with 10 mM HMA vesicular carrier. Average linear flow rate, 4.14 × 10-3 m s-1 (20 µL min-1). Other conditions are the same as Figure 1. (a, b) Elution curves for pure anthracene and naphthalene, respectively. (c) Mixture of anthracene and naphthalene.

of examined flow rates, uav ) 2.12 × 10-4-1.70 × 10-3 m s-1. Only a diffusion peak appears as long as uav is kept lower than 8.49 × 10-4 m s-1. Although a small nondiffusion peak is detected for higher flow rates, a diffusion peak is still dominant. In contrast, a nondiffusion peak becomes a main component of an elution curve for p-chlorophenol or naphthol, and an obvious diffusion peak is not detected when uav increases up to 1.27 × 10-3 or 8.49 × 10-4 m s-1, respectively. These observations mainly reflect different partition natures of these solutes. The partition coefficients between water and the SDS micelle have been reported to be 40.7-44.7, 219-243, and 607 for phenol, p-chlorophenol, and naphthol, corresponding to R ) 0.51-0.53, 0.85-0.87, and 0.94, respectively.30,31 Horizontal lines in Figure 2A show R values for these phenolic solutes in 100 mM SDS solution. Comparison of the elution profiles depicted in Figure 5 with Figure 4 allows us to evaluate their D2 values in aqueous solutions. The transformation from a diffusion peak to an intermediate on occurs at uav (30) Okada, T. Anal. Sci. 1993, 9, 59. (31) Nakamura, K.; Hayashi, K.; Ueda, I.; Fujiwara, H. Chem. Pharm. Bull. 1995, 43, 369.

Figure 6. Changes in elution profiles for I- and NO3- with flow rate and their resolution with 10 mM HMA carrier. Average linear flow rate, (1) 6.22 × 10-4 (3 µL min-1), (2) 1.04 × 10-3 (5 µL min-1), (3) 2.07 × 10-3 (10 µL min-1), (4) 4.14 × 10-3 (20 µL min-1), and (5) 6.22 × 10-3 m s-1 (30 µL min-1). Average linear flow rate for the bottom, 4.14 × 10-3 m s-1. Sample concentration, 0.1 mM for I- and 2 mM for NO3-. Arrows indicate peak edge discontinuities. Detection wavelength, 230 nm. Other conditions are the same as Figure 1.

) 8.49 × 10-4 m s-1 for phenol and at uav ) 4.25 × 10-4 m s-1 for chlorophenol and that from an intermediate peak to a nondiffusion peak takes place at uav ) 1.7 × 10-3 m s-1 for the latter. These transformations can be explained by assuming D2 )1 × 10-9 m2 s-1 (curves 2 and 2′ in Figure 2A), which is almost equal to the reported value for phenol (8.9 × 10-10 m2 s-1).32 Figures 2 and 3 strongly suggest that resolution of phenols be feasible simply by passing them through an open capillary. (32) CRC handbook of chemistry and physics, 81st ed.; Weast, R. C., Astle M. J., Beyer, W. H., Eds.; CRC Press: Cleveland, OH, 2000-2001.

Analytical Chemistry, Vol. 77, No. 18, September 15, 2005

6045

Figure 7. Resolution of pyrene and I- with 10 mM HMA and 10 mM NaBr carrier. Average linear flow rate, 2.07 × 10-3 m s-1 (10 µL min-1). Detection wavelength, 240 nm. Other conditions are the same as Figure 1.

Phenol, for example, can be resolved from naphthol with the 100 mM SDS carrier at uav ) 8.49 × 10-4 m s-1 as shown in Figure 4; phenol behaves as an almost diffusive solute, while naphthol is nondiffusive under this condition. Thus, the incorporation of molecular aggregates in the carrier solution allows the resolution of solutes by their simple passage through an open capillary, even when they are small dissolved molecules having similar diffusivity. HMA vesicular carriers were also examined as the carrier for separating small dissolved molecules. Quantitative discussion is not possible because no partition data are available for this vesicular system. Some examples for solute resolution are therefore shown in Figures 5-7. The elution behavior of aromatic hydrocarbons, anthracene and naphthalene, is shown in Figure 5. Anthracene and pyrene showed the identical elution profiles with any concentrations of HMA and flow rates, indicating that both compounds are completely partitioned into the vesicles. In contrast, naphthalene gave diffusive peaks over the uav range 8.49 × 10-4 -8.49 × 10-3 m s-1. Thus, as illustrated in Figure 5, naphthalene is resolved from anthracene by adjusting experimental conditions. The elution behavior of anions (I- and NO3-) was also studied with the HMA vesicular carriers. This system can be regarded as a kind of ion exchange. Less hydrated anions are, in general, more strongly bound by an anion exchanger. Figure 6 shows the elution curves of I- and NO3- with the HMA vesicular carrier. As expected from their anion-exchange properties, I- more strongly interacts with the cationic vesicles than NO3-. This is well reflected in their elution peak profiles; a clear nondiffusion peak is detected for I- at lower flow rate. It should be noted that NO3- shows

6046

Analytical Chemistry, Vol. 77, No. 18, September 15, 2005

unusual peak shapes, which appear to be produced by a superimposition of two different elution curves. Although this strange peak shape is similar to the intermediate peak profile, discontinuous peak edges were not found for other intermediate peaks. The slow kinetics of the release of anions accommodated in the inner phase of the vesicle may be involved in this strange peak emergence, albeit the origin of this strange peak shape has not been well elucidated at the present stage. The addition of an electrolyte into the carrier solution is another option in ionexchange-type separation mechanisms. The addition of NaBr to the HMA carrier, for example, makes I- more diffusive even at higher flow rates. This allows resolution between well-partitioned solutes such as I- and pyrene as depicted in Figure 7; the former is excluded from the vesicles by mass action, which does not affect the partition of the latter. In conclusion, the incorporation of molecular aggregates, which have much smaller diffusion properties than dissolved solutes, into the carrier solution in wide-bore capillary hydrodynamic chromatography has allowed resolution of dissolved molecules simply by passing them through an open capillary. The basic knowledge of separation mechanisms as well as the partition and diffusion data of solutes has facilitated the prediction and optimization of peak resolution. It should be noted that peak resolution is possible without any physical field or chromatographic stationary phase, albeit separation performance of the present method is highly restricted. However, we have confirmed by simulation that some improvements of separation performance are possible though some of these may be difficult to realize experimentally; e.g., (1) when the sample is injected at the center of the capillary, the peak broadening of a poorly diffusive solute can be substantially reduced and resolution becomes better; (2) if the location of injected sample is varied, the elution order and peak apex can be modified. Thus, though the separation performance of the present principle is highly restricted, it provides high potentials and new options in studying fundamental chemistry and can possibly substitute for existing methods in some applications. Finally, it should be kept in mind that the present phenomena may occur in any system where a laminar flow is utilized and must be taken into consideration in some cases. ACKNOWLEDGMENT This work has been in part supported by a Grant-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science, and Technology, Japan. Received for review April 25, 2005. Accepted July 17, 2005. AC050715Q